Skia merge (revision 3022)

This CL has companion changes to account for API updates in...
 (1) frameworks/base
 (2) external/webkit

Change-Id: Ibb989e76e8bd24313849f9631dbef42cdef9eb7d

32 files changed, 7368 insertions, 0 deletions

diff --git a/third_party/glu/LICENSE.txt b/third_party/glu/LICENSE.txt
new file mode 100644
index 0000000..e58ae38
--- /dev/null
+++ b/third_party/glu/LICENSE.txt
@@ -0,0 +1,31 @@
+SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008) 
+
+Copyright (C) [dates of first publication] Silicon Graphics, Inc. All
+Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice including the dates of first publication
+and either this permission notice or a reference to HYPERLINK
+"http://oss.sgi.com/projects/FreeB/"http://oss.sgi.com/projects/FreeB/
+shall be included in all copies or substantial portions of the
+Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL SILICON GRAPHICS, INC. BE LIABLE
+FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
+CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+Except as contained in this notice, the name of Silicon Graphics,
+Inc. shall not be used in advertising or otherwise to promote the
+sale, use or other dealings in this Software without prior written
+authorization from Silicon Graphics, Inc.
diff --git a/third_party/glu/README.skia b/third_party/glu/README.skia
new file mode 100644
index 0000000..0b6b624
--- /dev/null
+++ b/third_party/glu/README.skia
@@ -0,0 +1,41 @@
+This is a nearly verbatim copy of the GLU tessellator source code from
+SGI's OpenGL Sample Implementation at
+http://oss.sgi.com/projects/ogl-sample/ . Per
+http://oss.sgi.com/projects/FreeB/ , the code is covered under the SGI
+Free Software License B, version 2.0, a copy of which is in
+LICENSE.txt in this directory.
+
+The following changes were made in order to incorporate this code:
+
+  - The addition of a simplified gluos.h to eliminate operating system
+    dependencies.  The entry points to the tessellator were prefixed with
+    Sk_ to avoid symbol collisions with any host OS version of GLU via
+    #defines in gluos.h.
+
+  - The removal of inclusion of GL/glu.h and replacement with an
+    include of sk_glu.h.
+
+  - In tess.c, the obsolete entry points gluBeginPolygon,
+    gluNextContour and gluEndPolygon in tess.c were #if 0'd out.
+    Default branches were added to the switch statements in GotoState.
+
+  - In memalloc.h, the include of malloc.h was changed to an include
+    of stdlib.h.
+
+  - In normal.c, an unused variable "w" was removed from
+    __gl_projectPolygon. #if guards were placed around the definition
+    of the unused Normalize function.
+
+  - In priorityq-heap.c, an #include of <limits.h> was added.
+
+  - In sweep.c, IsWindingInside() was given a return value to silence a
+    warning-as-error in release builds.
+
+  - In sweep.c, DoneEdgeDict()'s fixedEdges was wrapped in #indef NDEBUG, to
+    silence a warning-as-error in release builds.
+
+  - In priorityq.c, render.c, and others:  the construct "if(1)...else" was
+  replaced with "do{...}while(1)" to silence a warning-as-error in Mac builds.
+
+  - rename all __gl_ functions to Sk__gl_ to avoid conflicting with other
+    static linkers of this library.
diff --git a/third_party/glu/gluos.h b/third_party/glu/gluos.h
new file mode 100644
index 0000000..5da46a5
--- /dev/null
+++ b/third_party/glu/gluos.h
@@ -0,0 +1,115 @@
+/*
+ * Copyright 2010, Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *     * Neither the name of Google Inc. nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef GLUOS_H_
+#define GLUOS_H_
+
+// This header provides the minimal definitions needed to compile the
+// GLU tessellator sources.
+#define GLAPIENTRY
+
+typedef unsigned char GLboolean;
+typedef double GLdouble;
+typedef unsigned int GLenum;
+typedef float GLfloat;
+typedef void GLvoid;
+
+#define gluNewTess Sk_gluNewTess
+#define gluDeleteTess Sk_gluDeleteTess
+#define gluTessProperty Sk_gluTessProperty
+#define gluGetTessProperty Sk_gluGetTessProperty
+#define gluTessNormal Sk_gluTessNormal
+#define gluTessCallback Sk_gluTessCallback
+#define gluTessVertex Sk_gluTessVertex
+#define gluTessBeginPolygon Sk_gluTessBeginPolygon
+#define gluTessBeginContour Sk_gluTessBeginContour
+#define gluTessEndContour Sk_gluTessEndContour
+#define gluTessEndPolygon Sk_gluTessEndPolygon
+
+#define __gl_noBeginData Sk__gl_noBeginData
+#define __gl_noEdgeFlagData Sk__gl_noEdgeFlagData
+#define __gl_noVertexData Sk__gl_noVertexData
+#define __gl_noEndData Sk__gl_noEndData
+#define __gl_noErrorData Sk__gl_noErrorData
+#define __gl_noCombineData Sk__gl_noCombineData
+#define __gl_computeInterior Sk__gl_computeInterior
+#define __gl_dictListDelete Sk__gl_dictListDelete
+#define __gl_dictListDeleteDict Sk__gl_dictListDeleteDict
+#define __gl_dictListInsertBefore Sk__gl_dictListInsertBefore
+#define __gl_dictListNewDict Sk__gl_dictListNewDict
+#define __gl_dictListSearch Sk__gl_dictListSearch
+#define __gl_edgeEval Sk__gl_edgeEval
+#define __gl_edgeIntersect Sk__gl_edgeIntersect
+#define __gl_edgeSign Sk__gl_edgeSign
+#define __gl_memInit Sk__gl_memInit
+#define __gl_meshAddEdgeVertex Sk__gl_meshAddEdgeVertex
+#define __gl_meshCheckMesh Sk__gl_meshCheckMesh
+#define __gl_meshConnect Sk__gl_meshConnect
+#define __gl_meshDelete Sk__gl_meshDelete
+#define __gl_meshDeleteMesh Sk__gl_meshDeleteMesh
+#define __gl_meshDiscardExterior Sk__gl_meshDiscardExterior
+#define __gl_meshMakeEdge Sk__gl_meshMakeEdge
+#define __gl_meshNewMesh Sk__gl_meshNewMesh
+#define __gl_meshSetWindingNumber Sk__gl_meshSetWindingNumber
+#define __gl_meshSplice Sk__gl_meshSplice
+#define __gl_meshSplitEdge Sk__gl_meshSplitEdge
+#define __gl_meshTessellateInterior Sk__gl_meshTessellateInterior
+#define __gl_meshTessellateMonoRegion Sk__gl_meshTessellateMonoRegion
+#define __gl_meshUnion Sk__gl_meshUnion
+#define __gl_meshZapFace Sk__gl_meshZapFace
+#define __gl_pqHeapDelete Sk__gl_pqHeapDelete
+#define __gl_pqHeapDeletePriorityQ Sk__gl_pqHeapDeletePriorityQ
+#define __gl_pqHeapExtractMin Sk__gl_pqHeapExtractMin
+#define __gl_pqHeapInit Sk__gl_pqHeapInit
+#define __gl_pqHeapInsert Sk__gl_pqHeapInsert
+#define __gl_pqHeapNewPriorityQ Sk__gl_pqHeapNewPriorityQ
+#define __gl_pqSortDelete Sk__gl_pqSortDelete
+#define __gl_pqSortDeletePriorityQ Sk__gl_pqSortDeletePriorityQ
+#define __gl_pqSortExtractMin Sk__gl_pqSortExtractMin
+#define __gl_pqSortInit Sk__gl_pqSortInit
+#define __gl_pqSortInsert Sk__gl_pqSortInsert
+#define __gl_pqSortIsEmpty Sk__gl_pqSortIsEmpty
+#define __gl_pqSortMinimum Sk__gl_pqSortMinimum
+#define __gl_pqSortNewPriorityQ Sk__gl_pqSortNewPriorityQ
+#define __gl_projectPolygon Sk__gl_projectPolygon
+#define __gl_renderBoundary Sk__gl_renderBoundary
+#define __gl_renderCache Sk__gl_renderCache
+#define __gl_renderMesh Sk__gl_renderMesh
+#define __gl_transEval Sk__gl_transEval
+#define __gl_transSign Sk__gl_transSign
+#define __gl_vertCCW Sk__gl_vertCCW
+#define __gl_vertLeq Sk__gl_vertLeq
+
+
+#undef MIN
+#undef MAX
+
+#endif  // GLUOS_H_
diff --git a/third_party/glu/libtess/GNUmakefile b/third_party/glu/libtess/GNUmakefile
new file mode 100644
index 0000000..0ae859f
--- /dev/null
+++ b/third_party/glu/libtess/GNUmakefile
@@ -0,0 +1,110 @@
+#!gmake
+#
+# License Applicability. Except to the extent portions of this file are
+# made subject to an alternative license as permitted in the SGI Free
+# Software License B, Version 1.1 (the "License"), the contents of this
+# file are subject only to the provisions of the License. You may not use
+# this file except in compliance with the License. You may obtain a copy
+# of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+# Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+#
+# http://oss.sgi.com/projects/FreeB
+#
+# Note that, as provided in the License, the Software is distributed on an
+# "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+# DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+# CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+# PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+#
+# Original Code. The Original Code is: OpenGL Sample Implementation,
+# Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+# Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+# Copyright in any portions created by third parties is as indicated
+# elsewhere herein. All Rights Reserved.
+#
+# Additional Notice Provisions: The application programming interfaces
+# established by SGI in conjunction with the Original Code are The
+# OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+# April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+# 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+# Window System(R) (Version 1.3), released October 19, 1998. This software
+# was created using the OpenGL(R) version 1.2.1 Sample Implementation
+# published by SGI, but has not been independently verified as being
+# compliant with the OpenGL(R) version 1.2.1 Specification.
+#
+# $Date$ $Revision$
+# $Header: //depot/main/gfx/lib/glu/libtess/GNUmakefile#9 $
+
+OBJECT_STYLE = N32
+
+include $(ROOT)/usr/include/make/commondefs
+
+TARGET = libtess.a
+
+TARGETS = $(TARGET)
+
+LCINCS = -I../include
+
+# Compilation flags:
+#
+# -DNDEBUG is for the production code; it removes all assertion checks
+#    (note that <assert.h> looks at this symbol).
+#
+# -DNO_MALLOPT uses regular malloc instead of the mallopt() version.
+#   ***** Unless you use this flag, you must use "-lmalloc" to link
+#   ***** your application!
+#
+# -DMEMORY_DEBUG turns on the M_DEBUG option of mallopt; this can
+#    increase the running time a LOT.
+#
+# -DGLU_TESS_API_FLOAT compiles a single-precision version of the library.
+#
+# -float prevents automatic promotion to double precision; this will produce
+#    faster code when compiled with -DGLU_TESS_API_FLOAT.
+#
+# -DNO_BRANCH_CONDITIONS uses & and | instead of && and || on a couple
+#    of heavily-used tests (VertEq and VertLeq); some compilers can generate
+#    better code with these (use special instructions to avoid branching).
+#
+# -DFOR_TRITE_TEST_PROGRAM is *only* for use with the test program called
+#    "trite".  It uses some variables which are defined by the test program,
+#    so you won't be able to link it with anything else.
+
+HFILES = \
+	dict.h \
+	dict-list.h \
+	geom.h \
+	memalloc.h \
+	mesh.h \
+	normal.h \
+	priorityq-heap.h \
+	priorityq-heap.c \
+	priorityq-sort.h \
+	priorityq.h \
+	render.h \
+	sweep.h \
+	tess.h \
+	tessmono.h \
+	$(NULL)
+
+CFILES = \
+	dict.c \
+	geom.c \
+	memalloc.c \
+	mesh.c \
+	normal.c \
+	priorityq.c \
+	render.c \
+	sweep.c \
+	tess.c \
+	tessmono.c \
+	$(NULL)
+
+default libs libs_install install: $(TARGET)
+
+headers headers_install apps:
+
+$(TARGET): $(OBJECTS)
+	$(AR) crl $@ $(OBJECTS);
+
+include $(COMMONRULES)
diff --git a/third_party/glu/libtess/Imakefile b/third_party/glu/libtess/Imakefile
new file mode 100644
index 0000000..fb99ce2
--- /dev/null
+++ b/third_party/glu/libtess/Imakefile
@@ -0,0 +1,61 @@
+XCOMM License Applicability. Except to the extent portions of this file are
+XCOMM made subject to an alternative license as permitted in the SGI Free
+XCOMM Software License B, Version 1.1 (the "License"), the contents of this
+XCOMM file are subject only to the provisions of the License. You may not use
+XCOMM this file except in compliance with the License. You may obtain a copy
+XCOMM of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+XCOMM Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+XCOMM 
+XCOMM http://oss.sgi.com/projects/FreeB
+XCOMM 
+XCOMM Note that, as provided in the License, the Software is distributed on an
+XCOMM "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+XCOMM DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+XCOMM CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+XCOMM PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+XCOMM 
+XCOMM Original Code. The Original Code is: OpenGL Sample Implementation,
+XCOMM Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+XCOMM Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+XCOMM Copyright in any portions created by third parties is as indicated
+XCOMM elsewhere herein. All Rights Reserved.
+XCOMM 
+XCOMM Additional Notice Provisions: The application programming interfaces
+XCOMM established by SGI in conjunction with the Original Code are The
+XCOMM OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+XCOMM April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+XCOMM 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+XCOMM Window System(R) (Version 1.3), released October 19, 1998. This software
+XCOMM was created using the OpenGL(R) version 1.2.1 Sample Implementation
+XCOMM published by SGI, but has not been independently verified as being
+XCOMM compliant with the OpenGL(R) version 1.2.1 Specification.
+XCOMM
+
+#include <Library.tmpl>
+
+OBJS = \
+	dict.o \
+	geom.o \
+	memalloc.o \
+	mesh.o \
+	normal.o \
+	priorityq.o \
+	render.o \
+	sweep.o \
+	tess.o \
+	tessmono.o
+
+INCLUDES = \
+	-I../include \
+	-I$(TOP)/include \
+	-I$(TOP)/include/GL
+
+DEFINES = \
+	-DNDEBUG
+
+NormalLibraryObjectRule()
+
+NormalLibraryTarget(tess, $(OBJS))
+
+DependTarget()
+CleanTarget()
diff --git a/third_party/glu/libtess/README b/third_party/glu/libtess/README
new file mode 100644
index 0000000..59f4ff2
--- /dev/null
+++ b/third_party/glu/libtess/README
@@ -0,0 +1,447 @@
+/*
+** $Header: /cvs/projects/ogl-sample/main/gfx/lib/glu/libtess/README,v 1.1 2000/04/26 05:53:59 ljp Exp $
+*/
+
+General Polygon Tesselation
+---------------------------
+
+  This note describes a tesselator for polygons consisting of one or
+  more closed contours.  It is backward-compatible with the current
+  OpenGL Utilities tesselator, and is intended to replace it.  Here is
+  a summary of the major differences:
+
+   - input contours can be intersecting, self-intersecting, or degenerate.
+  
+   - supports a choice of several winding rules for determining which parts
+     of the polygon are on the "interior".  This makes it possible to do
+     CSG operations on polygons.
+  
+   - boundary extraction: instead of tesselating the polygon, returns a
+     set of closed contours which separate the interior from the exterior.
+  
+   - returns the output as a small number of triangle fans and strips,
+     rather than a list of independent triangles (when possible).
+  
+   - output is available as an explicit mesh (a quad-edge structure),
+     in addition to the normal callback interface.
+  
+   - the algorithm used is extremely robust.
+
+
+The interface
+-------------
+
+  The tesselator state is maintained in a "tesselator object".
+  These are allocated and destroyed using
+
+     GLUtesselator *gluNewTess( void );
+     void gluDeleteTess( GLUtesselator *tess );
+
+  Several tesselator objects may be used simultaneously.
+
+  Inputs
+  ------
+  
+  The input contours are specified with the following routines:
+
+     void gluTessBeginPolygon( GLUtesselator *tess );
+     void gluTessBeginContour( GLUtesselator *tess );
+     void gluTessVertex( GLUtesselator *tess, GLUcoord coords[3], void *data );
+     void gluTessEndContour( GLUtesselator *tess );
+     void gluTessEndPolygon( GLUtesselator *tess );
+
+  Within each BeginPolygon/EndPolygon pair, there can be zero or more
+  calls to BeginContour/EndContour.  Within each contour, there are zero
+  or more calls to gluTessVertex().  The vertices specify a closed
+  contour (the last vertex of each contour is automatically linked to
+  the first).
+
+  "coords" give the coordinates of the vertex in 3-space.  For useful
+  results, all vertices should lie in some plane, since the vertices
+  are projected onto a plane before tesselation.  "data" is a pointer
+  to a user-defined vertex structure, which typically contains other
+  information such as color, texture coordinates, normal, etc.  It is
+  used to refer to the vertex during rendering.
+
+  The library can be compiled in single- or double-precision; the type
+  GLUcoord represents either "float" or "double" accordingly.  The GLU
+  version will be available in double-precision only.  Compile with
+  GLU_TESS_API_FLOAT defined to get the single-precision version.
+
+  When EndPolygon is called, the tesselation algorithm determines
+  which regions are interior to the given contours, according to one
+  of several "winding rules" described below.  The interior regions
+  are then tesselated, and the output is provided as callbacks.
+
+
+  Rendering Callbacks
+  -------------------
+
+  Callbacks are specified by the client using
+
+     void gluTessCallback( GLUtesselator *tess, GLenum which, void (*fn)());
+
+  If "fn" is NULL, any previously defined callback is discarded.
+  
+  The callbacks used to provide output are:	/* which == */
+
+     void begin( GLenum type );			/* GLU_TESS_BEGIN */
+     void edgeFlag( GLboolean flag );		/* GLU_TESS_EDGE_FLAG */
+     void vertex( void *data );			/* GLU_TESS_VERTEX */
+     void end( void );				/* GLU_TESS_END */
+
+  Any of the callbacks may be left undefined; if so, the corresponding
+  information will not be supplied during rendering.
+
+  The "begin" callback indicates the start of a primitive; type is one
+  of GL_TRIANGLE_STRIP, GL_TRIANGLE_FAN, or GL_TRIANGLES (but see the
+  notes on "boundary extraction" below).
+  
+  It is followed by any number of "vertex" callbacks, which supply the
+  vertices in the same order as expected by the corresponding glBegin()
+  call.  After the last vertex of a given primitive, there is a callback
+  to "end".
+
+  If the "edgeFlag" callback is provided, no triangle fans or strips
+  will be used.  When edgeFlag is called, if "flag" is GL_TRUE then each
+  vertex which follows begins an edge which lies on the polygon boundary
+  (ie. an edge which separates an interior region from an exterior one).
+  If "flag" is GL_FALSE, each vertex which follows begins an edge which lies
+  in the polygon interior.  "edgeFlag" will be called before the first
+  call to "vertex".
+
+  Other Callbacks
+  ---------------
+
+   void mesh( GLUmesh *mesh );			/* GLU_TESS_MESH */
+
+   - Returns an explicit mesh, represented using the quad-edge structure
+     (Guibas/Stolfi '85).  Other implementations of this interface might
+     use a different mesh structure, so this is available only only as an
+     SGI extension.  When the mesh is no longer needed, it should be freed
+     using
+
+	void gluDeleteMesh( GLUmesh *mesh );
+
+     There is a brief description of this data structure in the include
+     file "mesh.h".  For the full details, see L. Guibas and J. Stolfi,
+     Primitives for the manipulation of general subdivisions and the
+     computation of Voronoi diagrams, ACM Transactions on Graphics,
+     4(2):74-123, April 1985.  For an introduction, see the course notes
+     for CS348a, "Mathematical Foundations of Computer Graphics",
+     available at the Stanford bookstore (and taught during the fall
+     quarter).
+
+   void error( GLenum errno );			/* GLU_TESS_ERROR */
+
+   - errno is one of	GLU_TESS_MISSING_BEGIN_POLYGON,
+			GLU_TESS_MISSING_END_POLYGON,
+			GLU_TESS_MISSING_BEGIN_CONTOUR,
+			GLU_TESS_MISSING_END_CONTOUR,
+			GLU_TESS_COORD_TOO_LARGE,
+			GLU_TESS_NEED_COMBINE_CALLBACK
+
+     The first four are obvious.  The interface recovers from these
+     errors by inserting the missing call(s).
+  
+     GLU_TESS_COORD_TOO_LARGE says that some vertex coordinate exceeded
+     the predefined constant GLU_TESS_MAX_COORD in absolute value, and
+     that the value has been clamped.  (Coordinate values must be small
+     enough so that two can be multiplied together without overflow.)
+
+     GLU_TESS_NEED_COMBINE_CALLBACK says that the algorithm detected an
+     intersection between two edges in the input data, and the "combine"
+     callback (below) was not provided.  No output will be generated.
+
+
+   void combine( GLUcoord coords[3], void *data[4],	/* GLU_TESS_COMBINE */
+		 GLUcoord weight[4], void **outData );
+
+   - When the algorithm detects an intersection, or wishes to merge
+     features, it needs to create a new vertex.  The vertex is defined
+     as a linear combination of up to 4 existing vertices, referenced
+     by data[0..3].  The coefficients of the linear combination are
+     given by weight[0..3]; these weights always sum to 1.0.  All vertex
+     pointers are valid even when some of the weights are zero.
+     "coords" gives the location of the new vertex.
+
+     The user must allocate another vertex, interpolate parameters
+     using "data" and "weights", and return the new vertex pointer in
+     "outData".  This handle is supplied during rendering callbacks.
+     For example, if the polygon lies in an arbitrary plane in 3-space,
+     and we associate a color with each vertex, the combine callback might
+     look like this:
+    
+     void myCombine( GLUcoord coords[3], VERTEX *d[4],
+                     GLUcoord w[4], VERTEX **dataOut )
+     {
+        VERTEX *new = new_vertex();
+       
+        new->x = coords[0];
+        new->y = coords[1];
+        new->z = coords[2];
+        new->r = w[0]*d[0]->r + w[1]*d[1]->r + w[2]*d[2]->r + w[3]*d[3]->r;
+        new->g = w[0]*d[0]->g + w[1]*d[1]->g + w[2]*d[2]->g + w[3]*d[3]->g;
+        new->b = w[0]*d[0]->b + w[1]*d[1]->b + w[2]*d[2]->b + w[3]*d[3]->b;
+        new->a = w[0]*d[0]->a + w[1]*d[1]->a + w[2]*d[2]->a + w[3]*d[3]->a;
+        *dataOut = new;
+     }
+
+     If the algorithm detects an intersection, then the "combine" callback
+     must be defined, and must write a non-NULL pointer into "dataOut".
+     Otherwise the GLU_TESS_NEED_COMBINE_CALLBACK error occurs, and no
+     output is generated.  This is the only error that can occur during
+     tesselation and rendering.
+
+
+  Control over Tesselation
+  ------------------------
+  
+   void gluTessProperty( GLUtesselator *tess, GLenum which, GLUcoord value );
+
+   Properties defined:
+
+    - GLU_TESS_WINDING_RULE.  Possible values:
+
+	  GLU_TESS_WINDING_ODD
+	  GLU_TESS_WINDING_NONZERO
+	  GLU_TESS_WINDING_POSITIVE
+	  GLU_TESS_WINDING_NEGATIVE
+	  GLU_TESS_WINDING_ABS_GEQ_TWO
+
+      The input contours parition the plane into regions.  A winding
+      rule determines which of these regions are inside the polygon.
+      
+      For a single contour C, the winding number of a point x is simply
+      the signed number of revolutions we make around x as we travel
+      once around C (where CCW is positive).  When there are several
+      contours, the individual winding numbers are summed.  This
+      procedure associates a signed integer value with each point x in
+      the plane.  Note that the winding number is the same for all
+      points in a single region.
+
+      The winding rule classifies a region as "inside" if its winding
+      number belongs to the chosen category (odd, nonzero, positive,
+      negative, or absolute value of at least two).  The current GLU
+      tesselator implements the "odd" rule.  The "nonzero" rule is another
+      common way to define the interior.  The other three rules are
+      useful for polygon CSG operations (see below).
+
+    - GLU_TESS_BOUNDARY_ONLY.  Values: TRUE (non-zero) or FALSE (zero).
+
+      If TRUE, returns a set of closed contours which separate the
+      polygon interior and exterior (rather than a tesselation).
+      Exterior contours are oriented CCW with respect to the normal,
+      interior contours are oriented CW.  The GLU_TESS_BEGIN callback
+      uses the type GL_LINE_LOOP for each contour.
+      
+    - GLU_TESS_TOLERANCE.  Value: a real number between 0.0 and 1.0.
+
+      This specifies a tolerance for merging features to reduce the size
+      of the output.  For example, two vertices which are very close to
+      each other might be replaced by a single vertex.  The tolerance
+      is multiplied by the largest coordinate magnitude of any input vertex;
+      this specifies the maximum distance that any feature can move as the
+      result of a single merge operation.  If a single feature takes part
+      in several merge operations, the total distance moved could be larger.
+
+      Feature merging is completely optional; the tolerance is only a hint.
+      The implementation is free to merge in some cases and not in others,
+      or to never merge features at all.  The default tolerance is zero.
+      
+      The current implementation merges vertices only if they are exactly
+      coincident, regardless of the current tolerance.  A vertex is
+      spliced into an edge only if the implementation is unable to
+      distinguish which side of the edge the vertex lies on.
+      Two edges are merged only when both endpoints are identical.
+
+
+   void gluTessNormal( GLUtesselator *tess,
+		      GLUcoord x, GLUcoord y, GLUcoord z )
+
+    - Lets the user supply the polygon normal, if known.  All input data
+      is projected into a plane perpendicular to the normal before
+      tesselation.  All output triangles are oriented CCW with
+      respect to the normal (CW orientation can be obtained by
+      reversing the sign of the supplied normal).  For example, if
+      you know that all polygons lie in the x-y plane, call
+      "gluTessNormal(tess, 0.0, 0.0, 1.0)" before rendering any polygons.
+      
+    - If the supplied normal is (0,0,0) (the default value), the
+      normal is determined as follows.  The direction of the normal,
+      up to its sign, is found by fitting a plane to the vertices,
+      without regard to how the vertices are connected.  It is
+      expected that the input data lies approximately in plane;
+      otherwise projection perpendicular to the computed normal may
+      substantially change the geometry.  The sign of the normal is
+      chosen so that the sum of the signed areas of all input contours
+      is non-negative (where a CCW contour has positive area).
+    
+    - The supplied normal persists until it is changed by another
+      call to gluTessNormal.
+
+
+  Backward compatibility with the GLU tesselator
+  ----------------------------------------------
+
+  The preferred interface is the one described above.  The following
+  routines are obsolete, and are provided only for backward compatibility:
+
+    typedef GLUtesselator GLUtriangulatorObj;	/* obsolete name */
+
+    void gluBeginPolygon( GLUtesselator *tess );
+    void gluNextContour( GLUtesselator *tess, GLenum type );
+    void gluEndPolygon( GLUtesselator *tess );
+  
+  "type" is one of GLU_EXTERIOR, GLU_INTERIOR, GLU_CCW, GLU_CW, or
+  GLU_UNKNOWN.  It is ignored by the current GLU tesselator.
+  
+  GLU_BEGIN, GLU_VERTEX, GLU_END, GLU_ERROR, and GLU_EDGE_FLAG are defined
+  as synonyms for GLU_TESS_BEGIN, GLU_TESS_VERTEX, GLU_TESS_END,
+  GLU_TESS_ERROR, and GLU_TESS_EDGE_FLAG.
+
+
+Polygon CSG operations
+----------------------
+
+  The features of the tesselator make it easy to find the union, difference,
+  or intersection of several polygons.
+
+  First, assume that each polygon is defined so that the winding number
+  is 0 for each exterior region, and 1 for each interior region.  Under
+  this model, CCW contours define the outer boundary of the polygon, and
+  CW contours define holes.  Contours may be nested, but a nested
+  contour must be oriented oppositely from the contour that contains it.
+
+  If the original polygons do not satisfy this description, they can be
+  converted to this form by first running the tesselator with the
+  GLU_TESS_BOUNDARY_ONLY property turned on.  This returns a list of
+  contours satisfying the restriction above.  By allocating two
+  tesselator objects, the callbacks from one tesselator can be fed
+  directly to the input of another.
+
+  Given two or more polygons of the form above, CSG operations can be
+  implemented as follows:
+
+  Union
+     Draw all the input contours as a single polygon.  The winding number
+     of each resulting region is the number of original polygons
+     which cover it.  The union can be extracted using the
+     GLU_TESS_WINDING_NONZERO or GLU_TESS_WINDING_POSITIVE winding rules.
+     Note that with the nonzero rule, we would get the same result if
+     all contour orientations were reversed.
+
+  Intersection (two polygons at a time only)
+     Draw a single polygon using the contours from both input polygons.
+     Extract the result using GLU_TESS_WINDING_ABS_GEQ_TWO.  (Since this
+     winding rule looks at the absolute value, reversing all contour
+     orientations does not change the result.)
+
+  Difference
+  
+     Suppose we want to compute A \ (B union C union D).  Draw a single
+     polygon consisting of the unmodified contours from A, followed by
+     the contours of B,C,D with the vertex order reversed (this changes
+     the winding number of the interior regions to -1).  To extract the
+     result, use the GLU_TESS_WINDING_POSITIVE rule.
+   
+     If B,C,D are the result of a GLU_TESS_BOUNDARY_ONLY call, an
+     alternative to reversing the vertex order is to reverse the sign of
+     the supplied normal.  For example in the x-y plane, call
+     gluTessNormal( tess, 0.0, 0.0, -1.0 ).
+ 
+
+Performance
+-----------
+
+  The tesselator is not intended for immediate-mode rendering; when
+  possible the output should be cached in a user structure or display
+  list.  General polygon tesselation is an inherently difficult problem,
+  especially given the goal of extreme robustness.
+
+  The implementation makes an effort to output a small number of fans
+  and strips; this should improve the rendering performance when the
+  output is used in a display list.
+
+  Single-contour input polygons are first tested to see whether they can
+  be rendered as a triangle fan with respect to the first vertex (to
+  avoid running the full decomposition algorithm on convex polygons).
+  Non-convex polygons may be rendered by this "fast path" as well, if
+  the algorithm gets lucky in its choice of a starting vertex.
+
+  For best performance follow these guidelines:
+
+   - supply the polygon normal, if available, using gluTessNormal().
+     This represents about 10% of the computation time.  For example,
+     if all polygons lie in the x-y plane, use gluTessNormal(tess,0,0,1).
+
+   - render many polygons using the same tesselator object, rather than
+     allocating a new tesselator for each one.  (In a multi-threaded,
+     multi-processor environment you may get better performance using
+     several tesselators.)
+
+
+Comparison with the GLU tesselator
+----------------------------------
+
+  On polygons which make it through the "fast path", the tesselator is
+  3 to 5 times faster than the GLU tesselator.
+
+  On polygons which don't make it through the fast path (but which don't
+  have self-intersections or degeneracies), it is about 2 times slower.
+
+  On polygons with self-intersections or degeneraces, there is nothing
+  to compare against.
+
+  The new tesselator generates many more fans and strips, reducing the
+  number of vertices that need to be sent to the hardware.
+
+  Key to the statistics:
+
+	vert		number of input vertices on all contours
+	cntr		number of input contours
+	tri		number of triangles in all output primitives
+	strip		number of triangle strips
+	fan		number of triangle fans
+	ind		number of independent triangles
+	ms		number of milliseconds for tesselation
+			(on a 150MHz R4400 Indy)
+
+  Convex polygon examples:
+
+New:     3 vert,   1 cntr,     1 tri,   0 strip,   0 fan,     1 ind,  0.0459 ms
+Old:     3 vert,   1 cntr,     1 tri,   0 strip,   0 fan,     1 ind,   0.149 ms
+New:     4 vert,   1 cntr,     2 tri,   0 strip,   1 fan,     0 ind,  0.0459 ms
+Old:     4 vert,   1 cntr,     2 tri,   0 strip,   0 fan,     2 ind,   0.161 ms
+New:    36 vert,   1 cntr,    34 tri,   0 strip,   1 fan,     0 ind,   0.153 ms
+Old:    36 vert,   1 cntr,    34 tri,   0 strip,   0 fan,    34 ind,   0.621 ms
+
+  Concave single-contour polygons:
+
+New:     5 vert,   1 cntr,     3 tri,   0 strip,   1 fan,     0 ind,   0.052 ms
+Old:     5 vert,   1 cntr,     3 tri,   0 strip,   0 fan,     3 ind,   0.252 ms
+New:    19 vert,   1 cntr,    17 tri,   2 strip,   2 fan,     1 ind,   0.911 ms
+Old:    19 vert,   1 cntr,    17 tri,   0 strip,   0 fan,    17 ind,   0.529 ms
+New:   151 vert,   1 cntr,   149 tri,  13 strip,  18 fan,     3 ind,    6.82 ms
+Old:   151 vert,   1 cntr,   149 tri,   0 strip,   3 fan,   143 ind,     2.7 ms
+New:   574 vert,   1 cntr,   572 tri,  59 strip,  54 fan,    11 ind,    26.6 ms
+Old:   574 vert,   1 cntr,   572 tri,   0 strip,  31 fan,   499 ind,    12.4 ms
+
+  Multiple contours, but no intersections:
+
+New:     7 vert,   2 cntr,     7 tri,   1 strip,   0 fan,     0 ind,   0.527 ms
+Old:     7 vert,   2 cntr,     7 tri,   0 strip,   0 fan,     7 ind,   0.274 ms
+New:    81 vert,   6 cntr,    89 tri,   9 strip,   7 fan,     6 ind,    3.88 ms
+Old:    81 vert,   6 cntr,    89 tri,   0 strip,  13 fan,    61 ind,     2.2 ms
+New:   391 vert,  19 cntr,   413 tri,  37 strip,  32 fan,    26 ind,    20.2 ms
+Old:   391 vert,  19 cntr,   413 tri,   0 strip,  25 fan,   363 ind,    8.68 ms
+
+  Self-intersecting and degenerate examples:
+
+Bowtie:  4 vert,   1 cntr,     2 tri,   0 strip,   0 fan,     2 ind,   0.483 ms
+Star:    5 vert,   1 cntr,     5 tri,   0 strip,   0 fan,     5 ind,    0.91 ms
+Random: 24 vert,   7 cntr,    46 tri,   2 strip,  12 fan,     7 ind,    5.32 ms
+Font:  333 vert,   2 cntr,   331 tri,  32 strip,  16 fan,     3 ind,    14.1 ms
+:      167 vert,  35 cntr,   254 tri,   8 strip,  56 fan,    52 ind,    46.3 ms
+:       78 vert,   1 cntr,  2675 tri, 148 strip, 207 fan,   180 ind,     243 ms
+:    12480 vert,   2 cntr, 12478 tri, 736 strip,1275 fan,     5 ind,    1010 ms
diff --git a/third_party/glu/libtess/alg-outline b/third_party/glu/libtess/alg-outline
new file mode 100644
index 0000000..91ed7f3
--- /dev/null
+++ b/third_party/glu/libtess/alg-outline
@@ -0,0 +1,229 @@
+/*
+** $Header: /cvs/projects/ogl-sample/main/gfx/lib/glu/libtess/alg-outline,v 1.1 2000/04/26 05:53:59 ljp Exp $
+*/
+
+This is only a very brief overview.  There is quite a bit of
+additional documentation in the source code itself.
+
+
+Goals of robust tesselation
+---------------------------
+
+The tesselation algorithm is fundamentally a 2D algorithm.  We
+initially project all data into a plane; our goal is to robustly
+tesselate the projected data.  The same topological tesselation is
+then applied to the input data.
+
+Topologically, the output should always be a tesselation.  If the
+input is even slightly non-planar, then some triangles will
+necessarily be back-facing when viewed from some angles, but the goal
+is to minimize this effect.
+
+The algorithm needs some capability of cleaning up the input data as
+well as the numerical errors in its own calculations.  One way to do
+this is to specify a tolerance as defined above, and clean up the
+input and output during the line sweep process.  At the very least,
+the algorithm must handle coincident vertices, vertices incident to an
+edge, and coincident edges.
+
+
+Phases of the algorithm
+-----------------------
+
+1. Find the polygon normal N.
+2. Project the vertex data onto a plane.  It does not need to be
+   perpendicular to the normal, eg. we can project onto the plane
+   perpendicular to the coordinate axis whose dot product with N
+   is largest.
+3. Using a line-sweep algorithm, partition the plane into x-monotone
+   regions.  Any vertical line intersects an x-monotone region in
+   at most one interval.
+4. Triangulate the x-monotone regions.
+5. Group the triangles into strips and fans.
+
+
+Finding the normal vector
+-------------------------
+
+A common way to find a polygon normal is to compute the signed area
+when the polygon is projected along the three coordinate axes.  We
+can't do this, since contours can have zero area without being
+degenerate (eg. a bowtie).
+
+We fit a plane to the vertex data, ignoring how they are connected
+into contours.  Ideally this would be a least-squares fit; however for
+our purpose the accuracy of the normal is not important.  Instead we
+find three vertices which are widely separated, and compute the normal
+to the triangle they form.  The vertices are chosen so that the
+triangle has an area at least 1/sqrt(3) times the largest area of any
+triangle formed using the input vertices.  
+
+The contours do affect the orientation of the normal; after computing
+the normal, we check that the sum of the signed contour areas is
+non-negative, and reverse the normal if necessary.
+
+
+Projecting the vertices
+-----------------------
+
+We project the vertices onto a plane perpendicular to one of the three
+coordinate axes.  This helps numerical accuracy by removing a
+transformation step between the original input data and the data
+processed by the algorithm.  The projection also compresses the input
+data; the 2D distance between vertices after projection may be smaller
+than the original 2D distance.  However by choosing the coordinate
+axis whose dot product with the normal is greatest, the compression
+factor is at most 1/sqrt(3).
+
+Even though the *accuracy* of the normal is not that important (since
+we are projecting perpendicular to a coordinate axis anyway), the
+*robustness* of the computation is important.  For example, if there
+are many vertices which lie almost along a line, and one vertex V
+which is well-separated from the line, then our normal computation
+should involve V otherwise the results will be garbage.
+
+The advantage of projecting perpendicular to the polygon normal is
+that computed intersection points will be as close as possible to
+their ideal locations.  To get this behavior, define TRUE_PROJECT.
+
+
+The Line Sweep
+--------------
+
+There are three data structures: the mesh, the event queue, and the
+edge dictionary.
+
+The mesh is a "quad-edge" data structure which records the topology of
+the current decomposition; for details see the include file "mesh.h".
+
+The event queue simply holds all vertices (both original and computed
+ones), organized so that we can quickly extract the vertex with the
+minimum x-coord (and among those, the one with the minimum y-coord).
+
+The edge dictionary describes the current intersection of the sweep
+line with the regions of the polygon.  This is just an ordering of the
+edges which intersect the sweep line, sorted by their current order of
+intersection.  For each pair of edges, we store some information about
+the monotone region between them -- these are call "active regions"
+(since they are crossed by the current sweep line).
+
+The basic algorithm is to sweep from left to right, processing each
+vertex.  The processed portion of the mesh (left of the sweep line) is
+a planar decomposition.  As we cross each vertex, we update the mesh
+and the edge dictionary, then we check any newly adjacent pairs of
+edges to see if they intersect.
+
+A vertex can have any number of edges.  Vertices with many edges can
+be created as vertices are merged and intersection points are
+computed.  For unprocessed vertices (right of the sweep line), these
+edges are in no particular order around the vertex; for processed
+vertices, the topological ordering should match the geometric ordering.
+
+The vertex processing happens in two phases: first we process are the
+left-going edges (all these edges are currently in the edge
+dictionary).  This involves:
+
+ - deleting the left-going edges from the dictionary;
+ - relinking the mesh if necessary, so that the order of these edges around
+   the event vertex matches the order in the dictionary;
+ - marking any terminated regions (regions which lie between two left-going
+   edges) as either "inside" or "outside" according to their winding number.
+
+When there are no left-going edges, and the event vertex is in an
+"interior" region, we need to add an edge (to split the region into
+monotone pieces).  To do this we simply join the event vertex to the
+rightmost left endpoint of the upper or lower edge of the containing
+region.
+
+Then we process the right-going edges.  This involves:
+
+ - inserting the edges in the edge dictionary;
+ - computing the winding number of any newly created active regions.
+   We can compute this incrementally using the winding of each edge
+   that we cross as we walk through the dictionary.
+ - relinking the mesh if necessary, so that the order of these edges around
+   the event vertex matches the order in the dictionary;
+ - checking any newly adjacent edges for intersection and/or merging.
+
+If there are no right-going edges, again we need to add one to split
+the containing region into monotone pieces.  In our case it is most
+convenient to add an edge to the leftmost right endpoint of either
+containing edge; however we may need to change this later (see the
+code for details).
+
+
+Invariants
+----------
+
+These are the most important invariants maintained during the sweep.
+We define a function VertLeq(v1,v2) which defines the order in which
+vertices cross the sweep line, and a function EdgeLeq(e1,e2; loc)
+which says whether e1 is below e2 at the sweep event location "loc".
+This function is defined only at sweep event locations which lie
+between the rightmost left endpoint of {e1,e2}, and the leftmost right
+endpoint of {e1,e2}.
+
+Invariants for the Edge Dictionary.
+
+ - Each pair of adjacent edges e2=Succ(e1) satisfies EdgeLeq(e1,e2)
+   at any valid location of the sweep event.
+ - If EdgeLeq(e2,e1) as well (at any valid sweep event), then e1 and e2
+   share a common endpoint.
+ - For each e in the dictionary, e->Dst has been processed but not e->Org.
+ - Each edge e satisfies VertLeq(e->Dst,event) && VertLeq(event,e->Org)
+   where "event" is the current sweep line event.
+ - No edge e has zero length.
+ - No two edges have identical left and right endpoints.
+ 
+Invariants for the Mesh (the processed portion).
+
+ - The portion of the mesh left of the sweep line is a planar graph,
+   ie. there is *some* way to embed it in the plane.
+ - No processed edge has zero length.
+ - No two processed vertices have identical coordinates.
+ - Each "inside" region is monotone, ie. can be broken into two chains
+   of monotonically increasing vertices according to VertLeq(v1,v2)
+   - a non-invariant: these chains may intersect (slightly) due to
+     numerical errors, but this does not affect the algorithm's operation.
+
+Invariants for the Sweep.
+
+ - If a vertex has any left-going edges, then these must be in the edge
+   dictionary at the time the vertex is processed.
+ - If an edge is marked "fixUpperEdge" (it is a temporary edge introduced
+   by ConnectRightVertex), then it is the only right-going edge from
+   its associated vertex.  (This says that these edges exist only
+   when it is necessary.)
+
+
+Robustness
+----------
+
+The key to the robustness of the algorithm is maintaining the
+invariants above, especially the correct ordering of the edge
+dictionary.  We achieve this by:
+
+  1. Writing the numerical computations for maximum precision rather
+     than maximum speed.
+     
+  2. Making no assumptions at all about the results of the edge
+     intersection calculations -- for sufficiently degenerate inputs,
+     the computed location is not much better than a random number.
+     
+  3. When numerical errors violate the invariants, restore them
+     by making *topological* changes when necessary (ie. relinking
+     the mesh structure).
+     
+     
+Triangulation and Grouping
+--------------------------
+
+We finish the line sweep before doing any triangulation.  This is
+because even after a monotone region is complete, there can be further
+changes to its vertex data because of further vertex merging.
+
+After triangulating all monotone regions, we want to group the
+triangles into fans and strips.  We do this using a greedy approach.
+The triangulation itself is not optimized to reduce the number of
+primitives; we just try to get a reasonable decomposition of the
+computed triangulation.
diff --git a/third_party/glu/libtess/dict-list.h b/third_party/glu/libtess/dict-list.h
new file mode 100644
index 0000000..9c67c49
--- /dev/null
+++ b/third_party/glu/libtess/dict-list.h
@@ -0,0 +1,107 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/dict-list.h#5 $
+*/
+
+#ifndef __dict_list_h_
+#define __dict_list_h_
+
+/* Use #define's so that another heap implementation can use this one */
+
+#define DictKey		DictListKey
+#define Dict		DictList
+#define DictNode	DictListNode
+
+#define dictNewDict(frame,leq)		__gl_dictListNewDict(frame,leq)
+#define dictDeleteDict(dict)		__gl_dictListDeleteDict(dict)
+
+#define dictSearch(dict,key)		__gl_dictListSearch(dict,key)
+#define dictInsert(dict,key)		__gl_dictListInsert(dict,key)
+#define dictInsertBefore(dict,node,key)	__gl_dictListInsertBefore(dict,node,key)
+#define dictDelete(dict,node)		__gl_dictListDelete(dict,node)
+
+#define dictKey(n)			__gl_dictListKey(n)
+#define dictSucc(n)			__gl_dictListSucc(n)
+#define dictPred(n)			__gl_dictListPred(n)
+#define dictMin(d)			__gl_dictListMin(d)
+#define dictMax(d)			__gl_dictListMax(d)
+
+
+
+typedef void *DictKey;
+typedef struct Dict Dict;
+typedef struct DictNode DictNode;
+
+Dict		*dictNewDict(
+			void *frame,
+			int (*leq)(void *frame, DictKey key1, DictKey key2) );
+			
+void		dictDeleteDict( Dict *dict );
+
+/* Search returns the node with the smallest key greater than or equal
+ * to the given key.  If there is no such key, returns a node whose
+ * key is NULL.  Similarly, Succ(Max(d)) has a NULL key, etc.
+ */
+DictNode	*dictSearch( Dict *dict, DictKey key );
+DictNode	*dictInsertBefore( Dict *dict, DictNode *node, DictKey key );
+void		dictDelete( Dict *dict, DictNode *node );
+
+#define		__gl_dictListKey(n)	((n)->key)
+#define		__gl_dictListSucc(n)	((n)->next)
+#define		__gl_dictListPred(n)	((n)->prev)
+#define		__gl_dictListMin(d)	((d)->head.next)
+#define		__gl_dictListMax(d)	((d)->head.prev)
+#define	       __gl_dictListInsert(d,k) (dictInsertBefore((d),&(d)->head,(k)))
+
+
+/*** Private data structures ***/
+
+struct DictNode {
+  DictKey	key;
+  DictNode	*next;
+  DictNode	*prev;
+};
+
+struct Dict {
+  DictNode	head;
+  void		*frame;
+  int		(*leq)(void *frame, DictKey key1, DictKey key2);
+};
+
+#endif
diff --git a/third_party/glu/libtess/dict.c b/third_party/glu/libtess/dict.c
new file mode 100644
index 0000000..e149fd5
--- /dev/null
+++ b/third_party/glu/libtess/dict.c
@@ -0,0 +1,117 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/dict.c#5 $
+*/
+
+#include <stddef.h>
+#include "dict-list.h"
+#include "memalloc.h"
+
+/* really __gl_dictListNewDict */
+Dict *dictNewDict( void *frame,
+		   int (*leq)(void *frame, DictKey key1, DictKey key2) )
+{
+  Dict *dict = (Dict *) memAlloc( sizeof( Dict ));
+  DictNode *head;
+
+  if (dict == NULL) return NULL;
+
+  head = &dict->head;
+
+  head->key = NULL;
+  head->next = head;
+  head->prev = head;
+
+  dict->frame = frame;
+  dict->leq = leq;
+
+  return dict;
+}
+
+/* really __gl_dictListDeleteDict */
+void dictDeleteDict( Dict *dict )
+{
+  DictNode *node;
+
+  for( node = dict->head.next; node != &dict->head; node = node->next ) {
+    memFree( node );
+  }
+  memFree( dict );
+}
+
+/* really __gl_dictListInsertBefore */
+DictNode *dictInsertBefore( Dict *dict, DictNode *node, DictKey key )
+{
+  DictNode *newNode;
+
+  do {
+    node = node->prev;
+  } while( node->key != NULL && ! (*dict->leq)(dict->frame, node->key, key));
+
+  newNode = (DictNode *) memAlloc( sizeof( DictNode ));
+  if (newNode == NULL) return NULL;
+
+  newNode->key = key;
+  newNode->next = node->next;
+  node->next->prev = newNode;
+  newNode->prev = node;
+  node->next = newNode;
+
+  return newNode;
+}
+
+/* really __gl_dictListDelete */
+void dictDelete( Dict *dict, DictNode *node ) /*ARGSUSED*/
+{
+  node->next->prev = node->prev;
+  node->prev->next = node->next;
+  memFree( node );
+}
+
+/* really __gl_dictListSearch */
+DictNode *dictSearch( Dict *dict, DictKey key )
+{
+  DictNode *node = &dict->head;
+
+  do {
+    node = node->next;
+  } while( node->key != NULL && ! (*dict->leq)(dict->frame, key, node->key));
+
+  return node;
+}
diff --git a/third_party/glu/libtess/dict.h b/third_party/glu/libtess/dict.h
new file mode 100644
index 0000000..fee0769
--- /dev/null
+++ b/third_party/glu/libtess/dict.h
@@ -0,0 +1,109 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/dict.h#5 $
+*/
+
+#ifndef __dict_list_h_
+#define __dict_list_h_
+
+#include <sk_glu.h>
+
+/* Use #define's so that another heap implementation can use this one */
+
+#define DictKey		DictListKey
+#define Dict		DictList
+#define DictNode	DictListNode
+
+#define dictNewDict(frame,leq)		__gl_dictListNewDict(frame,leq)
+#define dictDeleteDict(dict)		__gl_dictListDeleteDict(dict)
+
+#define dictSearch(dict,key)		__gl_dictListSearch(dict,key)
+#define dictInsert(dict,key)		__gl_dictListInsert(dict,key)
+#define dictInsertBefore(dict,node,key)	__gl_dictListInsertBefore(dict,node,key)
+#define dictDelete(dict,node)		__gl_dictListDelete(dict,node)
+
+#define dictKey(n)			__gl_dictListKey(n)
+#define dictSucc(n)			__gl_dictListSucc(n)
+#define dictPred(n)			__gl_dictListPred(n)
+#define dictMin(d)			__gl_dictListMin(d)
+#define dictMax(d)			__gl_dictListMax(d)
+
+
+
+typedef void *DictKey;
+typedef struct Dict Dict;
+typedef struct DictNode DictNode;
+
+Dict		*dictNewDict(
+			void *frame,
+			int (*leq)(void *frame, DictKey key1, DictKey key2) );
+			
+void		dictDeleteDict( Dict *dict );
+
+/* Search returns the node with the smallest key greater than or equal
+ * to the given key.  If there is no such key, returns a node whose
+ * key is NULL.  Similarly, Succ(Max(d)) has a NULL key, etc.
+ */
+DictNode	*dictSearch( Dict *dict, DictKey key );
+DictNode	*dictInsertBefore( Dict *dict, DictNode *node, DictKey key );
+void		dictDelete( Dict *dict, DictNode *node );
+
+#define		__gl_dictListKey(n)	((n)->key)
+#define		__gl_dictListSucc(n)	((n)->next)
+#define		__gl_dictListPred(n)	((n)->prev)
+#define		__gl_dictListMin(d)	((d)->head.next)
+#define		__gl_dictListMax(d)	((d)->head.prev)
+#define	       __gl_dictListInsert(d,k) (dictInsertBefore((d),&(d)->head,(k)))
+
+
+/*** Private data structures ***/
+
+struct DictNode {
+  DictKey	key;
+  DictNode	*next;
+  DictNode	*prev;
+};
+
+struct Dict {
+  DictNode	head;
+  void		*frame;
+  int		(*leq)(void *frame, DictKey key1, DictKey key2);
+};
+
+#endif
diff --git a/third_party/glu/libtess/geom.c b/third_party/glu/libtess/geom.c
new file mode 100644
index 0000000..461d8b4
--- /dev/null
+++ b/third_party/glu/libtess/geom.c
@@ -0,0 +1,271 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/geom.c#5 $
+*/
+
+#include "gluos.h"
+#include <assert.h>
+#include "mesh.h"
+#include "geom.h"
+
+int __gl_vertLeq( GLUvertex *u, GLUvertex *v )
+{
+  /* Returns TRUE if u is lexicographically <= v. */
+
+  return VertLeq( u, v );
+}
+
+GLdouble __gl_edgeEval( GLUvertex *u, GLUvertex *v, GLUvertex *w )
+{
+  /* Given three vertices u,v,w such that VertLeq(u,v) && VertLeq(v,w),
+   * evaluates the t-coord of the edge uw at the s-coord of the vertex v.
+   * Returns v->t - (uw)(v->s), ie. the signed distance from uw to v.
+   * If uw is vertical (and thus passes thru v), the result is zero.
+   *
+   * The calculation is extremely accurate and stable, even when v
+   * is very close to u or w.  In particular if we set v->t = 0 and
+   * let r be the negated result (this evaluates (uw)(v->s)), then
+   * r is guaranteed to satisfy MIN(u->t,w->t) <= r <= MAX(u->t,w->t).
+   */
+  GLdouble gapL, gapR;
+
+  assert( VertLeq( u, v ) && VertLeq( v, w ));
+  
+  gapL = v->s - u->s;
+  gapR = w->s - v->s;
+
+  if( gapL + gapR > 0 ) {
+    if( gapL < gapR ) {
+      return (v->t - u->t) + (u->t - w->t) * (gapL / (gapL + gapR));
+    } else {
+      return (v->t - w->t) + (w->t - u->t) * (gapR / (gapL + gapR));
+    }
+  }
+  /* vertical line */
+  return 0;
+}
+
+GLdouble __gl_edgeSign( GLUvertex *u, GLUvertex *v, GLUvertex *w )
+{
+  /* Returns a number whose sign matches EdgeEval(u,v,w) but which
+   * is cheaper to evaluate.  Returns > 0, == 0 , or < 0
+   * as v is above, on, or below the edge uw.
+   */
+  GLdouble gapL, gapR;
+
+  assert( VertLeq( u, v ) && VertLeq( v, w ));
+  
+  gapL = v->s - u->s;
+  gapR = w->s - v->s;
+
+  if( gapL + gapR > 0 ) {
+    return (v->t - w->t) * gapL + (v->t - u->t) * gapR;
+  }
+  /* vertical line */
+  return 0;
+}
+
+
+/***********************************************************************
+ * Define versions of EdgeSign, EdgeEval with s and t transposed.
+ */
+
+GLdouble __gl_transEval( GLUvertex *u, GLUvertex *v, GLUvertex *w )
+{
+  /* Given three vertices u,v,w such that TransLeq(u,v) && TransLeq(v,w),
+   * evaluates the t-coord of the edge uw at the s-coord of the vertex v.
+   * Returns v->s - (uw)(v->t), ie. the signed distance from uw to v.
+   * If uw is vertical (and thus passes thru v), the result is zero.
+   *
+   * The calculation is extremely accurate and stable, even when v
+   * is very close to u or w.  In particular if we set v->s = 0 and
+   * let r be the negated result (this evaluates (uw)(v->t)), then
+   * r is guaranteed to satisfy MIN(u->s,w->s) <= r <= MAX(u->s,w->s).
+   */
+  GLdouble gapL, gapR;
+
+  assert( TransLeq( u, v ) && TransLeq( v, w ));
+  
+  gapL = v->t - u->t;
+  gapR = w->t - v->t;
+
+  if( gapL + gapR > 0 ) {
+    if( gapL < gapR ) {
+      return (v->s - u->s) + (u->s - w->s) * (gapL / (gapL + gapR));
+    } else {
+      return (v->s - w->s) + (w->s - u->s) * (gapR / (gapL + gapR));
+    }
+  }
+  /* vertical line */
+  return 0;
+}
+
+GLdouble __gl_transSign( GLUvertex *u, GLUvertex *v, GLUvertex *w )
+{
+  /* Returns a number whose sign matches TransEval(u,v,w) but which
+   * is cheaper to evaluate.  Returns > 0, == 0 , or < 0
+   * as v is above, on, or below the edge uw.
+   */
+  GLdouble gapL, gapR;
+
+  assert( TransLeq( u, v ) && TransLeq( v, w ));
+  
+  gapL = v->t - u->t;
+  gapR = w->t - v->t;
+
+  if( gapL + gapR > 0 ) {
+    return (v->s - w->s) * gapL + (v->s - u->s) * gapR;
+  }
+  /* vertical line */
+  return 0;
+}
+
+
+int __gl_vertCCW( GLUvertex *u, GLUvertex *v, GLUvertex *w )
+{
+  /* For almost-degenerate situations, the results are not reliable.
+   * Unless the floating-point arithmetic can be performed without
+   * rounding errors, *any* implementation will give incorrect results
+   * on some degenerate inputs, so the client must have some way to
+   * handle this situation.
+   */
+  return (u->s*(v->t - w->t) + v->s*(w->t - u->t) + w->s*(u->t - v->t)) >= 0;
+}
+
+/* Given parameters a,x,b,y returns the value (b*x+a*y)/(a+b),
+ * or (x+y)/2 if a==b==0.  It requires that a,b >= 0, and enforces
+ * this in the rare case that one argument is slightly negative.
+ * The implementation is extremely stable numerically.
+ * In particular it guarantees that the result r satisfies
+ * MIN(x,y) <= r <= MAX(x,y), and the results are very accurate
+ * even when a and b differ greatly in magnitude.
+ */
+#define RealInterpolate(a,x,b,y)			\
+  (a = (a < 0) ? 0 : a, b = (b < 0) ? 0 : b,		\
+  ((a <= b) ? ((b == 0) ? ((x+y) / 2)			\
+                        : (x + (y-x) * (a/(a+b))))	\
+            : (y + (x-y) * (b/(a+b)))))
+
+#ifndef FOR_TRITE_TEST_PROGRAM
+#define Interpolate(a,x,b,y)	RealInterpolate(a,x,b,y)
+#else
+
+/* Claim: the ONLY property the sweep algorithm relies on is that
+ * MIN(x,y) <= r <= MAX(x,y).  This is a nasty way to test that.
+ */
+#include <stdlib.h>
+extern int RandomInterpolate;
+
+GLdouble Interpolate( GLdouble a, GLdouble x, GLdouble b, GLdouble y)
+{
+printf("*********************%d\n",RandomInterpolate);
+  if( RandomInterpolate ) {
+    a = 1.2 * drand48() - 0.1;
+    a = (a < 0) ? 0 : ((a > 1) ? 1 : a);
+    b = 1.0 - a;
+  }
+  return RealInterpolate(a,x,b,y);
+}
+
+#endif
+
+#define Swap(a,b)	do { GLUvertex *t = a; a = b; b = t; } while(0)
+
+void __gl_edgeIntersect( GLUvertex *o1, GLUvertex *d1,
+			 GLUvertex *o2, GLUvertex *d2,
+			 GLUvertex *v )
+/* Given edges (o1,d1) and (o2,d2), compute their point of intersection.
+ * The computed point is guaranteed to lie in the intersection of the
+ * bounding rectangles defined by each edge.
+ */
+{
+  GLdouble z1, z2;
+
+  /* This is certainly not the most efficient way to find the intersection
+   * of two line segments, but it is very numerically stable.
+   *
+   * Strategy: find the two middle vertices in the VertLeq ordering,
+   * and interpolate the intersection s-value from these.  Then repeat
+   * using the TransLeq ordering to find the intersection t-value.
+   */
+
+  if( ! VertLeq( o1, d1 )) { Swap( o1, d1 ); }
+  if( ! VertLeq( o2, d2 )) { Swap( o2, d2 ); }
+  if( ! VertLeq( o1, o2 )) { Swap( o1, o2 ); Swap( d1, d2 ); }
+
+  if( ! VertLeq( o2, d1 )) {
+    /* Technically, no intersection -- do our best */
+    v->s = (o2->s + d1->s) / 2;
+  } else if( VertLeq( d1, d2 )) {
+    /* Interpolate between o2 and d1 */
+    z1 = EdgeEval( o1, o2, d1 );
+    z2 = EdgeEval( o2, d1, d2 );
+    if( z1+z2 < 0 ) { z1 = -z1; z2 = -z2; }
+    v->s = Interpolate( z1, o2->s, z2, d1->s );
+  } else {
+    /* Interpolate between o2 and d2 */
+    z1 = EdgeSign( o1, o2, d1 );
+    z2 = -EdgeSign( o1, d2, d1 );
+    if( z1+z2 < 0 ) { z1 = -z1; z2 = -z2; }
+    v->s = Interpolate( z1, o2->s, z2, d2->s );
+  }
+
+  /* Now repeat the process for t */
+
+  if( ! TransLeq( o1, d1 )) { Swap( o1, d1 ); }
+  if( ! TransLeq( o2, d2 )) { Swap( o2, d2 ); }
+  if( ! TransLeq( o1, o2 )) { Swap( o1, o2 ); Swap( d1, d2 ); }
+
+  if( ! TransLeq( o2, d1 )) {
+    /* Technically, no intersection -- do our best */
+    v->t = (o2->t + d1->t) / 2;
+  } else if( TransLeq( d1, d2 )) {
+    /* Interpolate between o2 and d1 */
+    z1 = TransEval( o1, o2, d1 );
+    z2 = TransEval( o2, d1, d2 );
+    if( z1+z2 < 0 ) { z1 = -z1; z2 = -z2; }
+    v->t = Interpolate( z1, o2->t, z2, d1->t );
+  } else {
+    /* Interpolate between o2 and d2 */
+    z1 = TransSign( o1, o2, d1 );
+    z2 = -TransSign( o1, d2, d1 );
+    if( z1+z2 < 0 ) { z1 = -z1; z2 = -z2; }
+    v->t = Interpolate( z1, o2->t, z2, d2->t );
+  }
+}
diff --git a/third_party/glu/libtess/geom.h b/third_party/glu/libtess/geom.h
new file mode 100644
index 0000000..b4d0c66
--- /dev/null
+++ b/third_party/glu/libtess/geom.h
@@ -0,0 +1,90 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/geom.h#5 $
+*/
+
+#ifndef __geom_h_
+#define __geom_h_
+
+#include "mesh.h"
+
+#ifdef NO_BRANCH_CONDITIONS
+/* MIPS architecture has special instructions to evaluate boolean
+ * conditions -- more efficient than branching, IF you can get the
+ * compiler to generate the right instructions (SGI compiler doesn't)
+ */
+#define VertEq(u,v)	(((u)->s == (v)->s) & ((u)->t == (v)->t))
+#define VertLeq(u,v)	(((u)->s < (v)->s) | \
+                         ((u)->s == (v)->s & (u)->t <= (v)->t))
+#else
+#define VertEq(u,v)	((u)->s == (v)->s && (u)->t == (v)->t)
+#define VertLeq(u,v)	(((u)->s < (v)->s) || \
+                         ((u)->s == (v)->s && (u)->t <= (v)->t))
+#endif
+
+#define EdgeEval(u,v,w)	__gl_edgeEval(u,v,w)
+#define EdgeSign(u,v,w)	__gl_edgeSign(u,v,w)
+
+/* Versions of VertLeq, EdgeSign, EdgeEval with s and t transposed. */
+
+#define TransLeq(u,v)	(((u)->t < (v)->t) || \
+                         ((u)->t == (v)->t && (u)->s <= (v)->s))
+#define TransEval(u,v,w)	__gl_transEval(u,v,w)
+#define TransSign(u,v,w)	__gl_transSign(u,v,w)
+
+
+#define EdgeGoesLeft(e)		VertLeq( (e)->Dst, (e)->Org )
+#define EdgeGoesRight(e)	VertLeq( (e)->Org, (e)->Dst )
+
+#define ABS(x)	((x) < 0 ? -(x) : (x))
+#define VertL1dist(u,v)	(ABS(u->s - v->s) + ABS(u->t - v->t))
+
+#define VertCCW(u,v,w)	__gl_vertCCW(u,v,w)
+
+int		__gl_vertLeq( GLUvertex *u, GLUvertex *v );
+GLdouble	__gl_edgeEval( GLUvertex *u, GLUvertex *v, GLUvertex *w );
+GLdouble	__gl_edgeSign( GLUvertex *u, GLUvertex *v, GLUvertex *w );
+GLdouble	__gl_transEval( GLUvertex *u, GLUvertex *v, GLUvertex *w );
+GLdouble	__gl_transSign( GLUvertex *u, GLUvertex *v, GLUvertex *w );
+int		__gl_vertCCW( GLUvertex *u, GLUvertex *v, GLUvertex *w );
+void		__gl_edgeIntersect( GLUvertex *o1, GLUvertex *d1,
+				    GLUvertex *o2, GLUvertex *d2,
+				    GLUvertex *v );
+
+#endif
diff --git a/third_party/glu/libtess/memalloc.c b/third_party/glu/libtess/memalloc.c
new file mode 100644
index 0000000..bd5b4ac
--- /dev/null
+++ b/third_party/glu/libtess/memalloc.c
@@ -0,0 +1,62 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/memalloc.c#5 $
+*/
+
+#include "memalloc.h"
+#include "string.h"
+
+int __gl_memInit( size_t maxFast )
+{
+#ifndef NO_MALLOPT
+/*  mallopt( M_MXFAST, maxFast );*/
+#ifdef MEMORY_DEBUG
+  mallopt( M_DEBUG, 1 );
+#endif
+#endif
+   return 1;
+}
+
+#ifdef MEMORY_DEBUG
+void *__gl_memAlloc( size_t n )
+{
+  return memset( malloc( n ), 0xa5, n );
+}
+#endif
+
diff --git a/third_party/glu/libtess/memalloc.h b/third_party/glu/libtess/memalloc.h
new file mode 100644
index 0000000..7133ada
--- /dev/null
+++ b/third_party/glu/libtess/memalloc.h
@@ -0,0 +1,62 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/memalloc.h#5 $
+*/
+
+#ifndef __memalloc_simple_h_
+#define __memalloc_simple_h_
+
+#include <sk_glu.h>
+#include <stdlib.h>
+
+#define memRealloc	realloc
+#define memFree		free
+
+#define memInit		__gl_memInit
+/*extern void		__gl_memInit( size_t );*/
+extern int		__gl_memInit( size_t );
+
+#ifndef MEMORY_DEBUG
+#define memAlloc	malloc
+#else
+#define memAlloc	__gl_memAlloc
+extern void *		__gl_memAlloc( size_t );
+#endif
+
+#endif
diff --git a/third_party/glu/libtess/mesh.c b/third_party/glu/libtess/mesh.c
new file mode 100644
index 0000000..7305fab
--- /dev/null
+++ b/third_party/glu/libtess/mesh.c
@@ -0,0 +1,796 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/mesh.c#6 $
+*/
+
+#include "gluos.h"
+#include <stddef.h>
+#include <assert.h>
+#include "mesh.h"
+#include "memalloc.h"
+
+#define TRUE 1
+#define FALSE 0
+
+static GLUvertex *allocVertex()
+{
+   return (GLUvertex *)memAlloc( sizeof( GLUvertex ));
+}
+
+static GLUface *allocFace()
+{
+   return (GLUface *)memAlloc( sizeof( GLUface ));
+}
+
+/************************ Utility Routines ************************/
+
+/* Allocate and free half-edges in pairs for efficiency.
+ * The *only* place that should use this fact is allocation/free.
+ */
+typedef struct { GLUhalfEdge e, eSym; } EdgePair;
+
+/* MakeEdge creates a new pair of half-edges which form their own loop.
+ * No vertex or face structures are allocated, but these must be assigned
+ * before the current edge operation is completed.
+ */
+static GLUhalfEdge *MakeEdge( GLUhalfEdge *eNext )
+{
+  GLUhalfEdge *e;
+  GLUhalfEdge *eSym;
+  GLUhalfEdge *ePrev;
+  EdgePair *pair = (EdgePair *)memAlloc( sizeof( EdgePair ));
+  if (pair == NULL) return NULL;
+
+  e = &pair->e;
+  eSym = &pair->eSym;
+
+  /* Make sure eNext points to the first edge of the edge pair */
+  if( eNext->Sym < eNext ) { eNext = eNext->Sym; }
+
+  /* Insert in circular doubly-linked list before eNext.
+   * Note that the prev pointer is stored in Sym->next.
+   */
+  ePrev = eNext->Sym->next;
+  eSym->next = ePrev;
+  ePrev->Sym->next = e;
+  e->next = eNext;
+  eNext->Sym->next = eSym;
+
+  e->Sym = eSym;
+  e->Onext = e;
+  e->Lnext = eSym;
+  e->Org = NULL;
+  e->Lface = NULL;
+  e->winding = 0;
+  e->activeRegion = NULL;
+
+  eSym->Sym = e;
+  eSym->Onext = eSym;
+  eSym->Lnext = e;
+  eSym->Org = NULL;
+  eSym->Lface = NULL;
+  eSym->winding = 0;
+  eSym->activeRegion = NULL;
+
+  return e;
+}
+
+/* Splice( a, b ) is best described by the Guibas/Stolfi paper or the
+ * CS348a notes (see mesh.h).  Basically it modifies the mesh so that
+ * a->Onext and b->Onext are exchanged.  This can have various effects
+ * depending on whether a and b belong to different face or vertex rings.
+ * For more explanation see __gl_meshSplice() below.
+ */
+static void Splice( GLUhalfEdge *a, GLUhalfEdge *b )
+{
+  GLUhalfEdge *aOnext = a->Onext;
+  GLUhalfEdge *bOnext = b->Onext;
+
+  aOnext->Sym->Lnext = b;
+  bOnext->Sym->Lnext = a;
+  a->Onext = bOnext;
+  b->Onext = aOnext;
+}
+
+/* MakeVertex( newVertex, eOrig, vNext ) attaches a new vertex and makes it the
+ * origin of all edges in the vertex loop to which eOrig belongs. "vNext" gives
+ * a place to insert the new vertex in the global vertex list.  We insert
+ * the new vertex *before* vNext so that algorithms which walk the vertex
+ * list will not see the newly created vertices.
+ */
+static void MakeVertex( GLUvertex *newVertex, 
+			GLUhalfEdge *eOrig, GLUvertex *vNext )
+{
+  GLUhalfEdge *e;
+  GLUvertex *vPrev;
+  GLUvertex *vNew = newVertex;
+
+  assert(vNew != NULL);
+
+  /* insert in circular doubly-linked list before vNext */
+  vPrev = vNext->prev;
+  vNew->prev = vPrev;
+  vPrev->next = vNew;
+  vNew->next = vNext;
+  vNext->prev = vNew;
+
+  vNew->anEdge = eOrig;
+  vNew->data = NULL;
+  /* leave coords, s, t undefined */
+
+  /* fix other edges on this vertex loop */
+  e = eOrig;
+  do {
+    e->Org = vNew;
+    e = e->Onext;
+  } while( e != eOrig );
+}
+
+/* MakeFace( newFace, eOrig, fNext ) attaches a new face and makes it the left
+ * face of all edges in the face loop to which eOrig belongs.  "fNext" gives
+ * a place to insert the new face in the global face list.  We insert
+ * the new face *before* fNext so that algorithms which walk the face
+ * list will not see the newly created faces.
+ */
+static void MakeFace( GLUface *newFace, GLUhalfEdge *eOrig, GLUface *fNext )
+{
+  GLUhalfEdge *e;
+  GLUface *fPrev;
+  GLUface *fNew = newFace;
+
+  assert(fNew != NULL); 
+
+  /* insert in circular doubly-linked list before fNext */
+  fPrev = fNext->prev;
+  fNew->prev = fPrev;
+  fPrev->next = fNew;
+  fNew->next = fNext;
+  fNext->prev = fNew;
+
+  fNew->anEdge = eOrig;
+  fNew->data = NULL;
+  fNew->trail = NULL;
+  fNew->marked = FALSE;
+
+  /* The new face is marked "inside" if the old one was.  This is a
+   * convenience for the common case where a face has been split in two.
+   */
+  fNew->inside = fNext->inside;
+
+  /* fix other edges on this face loop */
+  e = eOrig;
+  do {
+    e->Lface = fNew;
+    e = e->Lnext;
+  } while( e != eOrig );
+}
+
+/* KillEdge( eDel ) destroys an edge (the half-edges eDel and eDel->Sym),
+ * and removes from the global edge list.
+ */
+static void KillEdge( GLUhalfEdge *eDel )
+{
+  GLUhalfEdge *ePrev, *eNext;
+
+  /* Half-edges are allocated in pairs, see EdgePair above */
+  if( eDel->Sym < eDel ) { eDel = eDel->Sym; }
+
+  /* delete from circular doubly-linked list */
+  eNext = eDel->next;
+  ePrev = eDel->Sym->next;
+  eNext->Sym->next = ePrev;
+  ePrev->Sym->next = eNext;
+
+  memFree( eDel );
+}
+
+
+/* KillVertex( vDel ) destroys a vertex and removes it from the global
+ * vertex list.  It updates the vertex loop to point to a given new vertex.
+ */
+static void KillVertex( GLUvertex *vDel, GLUvertex *newOrg )
+{
+  GLUhalfEdge *e, *eStart = vDel->anEdge;
+  GLUvertex *vPrev, *vNext;
+
+  /* change the origin of all affected edges */
+  e = eStart;
+  do {
+    e->Org = newOrg;
+    e = e->Onext;
+  } while( e != eStart );
+
+  /* delete from circular doubly-linked list */
+  vPrev = vDel->prev;
+  vNext = vDel->next;
+  vNext->prev = vPrev;
+  vPrev->next = vNext;
+
+  memFree( vDel );
+}
+
+/* KillFace( fDel ) destroys a face and removes it from the global face
+ * list.  It updates the face loop to point to a given new face.
+ */
+static void KillFace( GLUface *fDel, GLUface *newLface )
+{
+  GLUhalfEdge *e, *eStart = fDel->anEdge;
+  GLUface *fPrev, *fNext;
+
+  /* change the left face of all affected edges */
+  e = eStart;
+  do {
+    e->Lface = newLface;
+    e = e->Lnext;
+  } while( e != eStart );
+
+  /* delete from circular doubly-linked list */
+  fPrev = fDel->prev;
+  fNext = fDel->next;
+  fNext->prev = fPrev;
+  fPrev->next = fNext;
+
+  memFree( fDel );
+}
+
+
+/****************** Basic Edge Operations **********************/
+
+/* __gl_meshMakeEdge creates one edge, two vertices, and a loop (face).
+ * The loop consists of the two new half-edges.
+ */
+GLUhalfEdge *__gl_meshMakeEdge( GLUmesh *mesh )
+{
+  GLUvertex *newVertex1= allocVertex();
+  GLUvertex *newVertex2= allocVertex();
+  GLUface *newFace= allocFace();
+  GLUhalfEdge *e;
+
+  /* if any one is null then all get freed */
+  if (newVertex1 == NULL || newVertex2 == NULL || newFace == NULL) {
+     if (newVertex1 != NULL) memFree(newVertex1);
+     if (newVertex2 != NULL) memFree(newVertex2);
+     if (newFace != NULL) memFree(newFace);     
+     return NULL;
+  } 
+
+  e = MakeEdge( &mesh->eHead );
+  if (e == NULL) return NULL;
+
+  MakeVertex( newVertex1, e, &mesh->vHead );
+  MakeVertex( newVertex2, e->Sym, &mesh->vHead );
+  MakeFace( newFace, e, &mesh->fHead );
+  return e;
+}
+  
+
+/* __gl_meshSplice( eOrg, eDst ) is the basic operation for changing the
+ * mesh connectivity and topology.  It changes the mesh so that
+ *	eOrg->Onext <- OLD( eDst->Onext )
+ *	eDst->Onext <- OLD( eOrg->Onext )
+ * where OLD(...) means the value before the meshSplice operation.
+ *
+ * This can have two effects on the vertex structure:
+ *  - if eOrg->Org != eDst->Org, the two vertices are merged together
+ *  - if eOrg->Org == eDst->Org, the origin is split into two vertices
+ * In both cases, eDst->Org is changed and eOrg->Org is untouched.
+ *
+ * Similarly (and independently) for the face structure,
+ *  - if eOrg->Lface == eDst->Lface, one loop is split into two
+ *  - if eOrg->Lface != eDst->Lface, two distinct loops are joined into one
+ * In both cases, eDst->Lface is changed and eOrg->Lface is unaffected.
+ *
+ * Some special cases:
+ * If eDst == eOrg, the operation has no effect.
+ * If eDst == eOrg->Lnext, the new face will have a single edge.
+ * If eDst == eOrg->Lprev, the old face will have a single edge.
+ * If eDst == eOrg->Onext, the new vertex will have a single edge.
+ * If eDst == eOrg->Oprev, the old vertex will have a single edge.
+ */
+int __gl_meshSplice( GLUhalfEdge *eOrg, GLUhalfEdge *eDst )
+{
+  int joiningLoops = FALSE;
+  int joiningVertices = FALSE;
+
+  if( eOrg == eDst ) return 1;
+
+  if( eDst->Org != eOrg->Org ) {
+    /* We are merging two disjoint vertices -- destroy eDst->Org */
+    joiningVertices = TRUE;
+    KillVertex( eDst->Org, eOrg->Org );
+  }
+  if( eDst->Lface != eOrg->Lface ) {
+    /* We are connecting two disjoint loops -- destroy eDst->Lface */
+    joiningLoops = TRUE;
+    KillFace( eDst->Lface, eOrg->Lface );
+  }
+
+  /* Change the edge structure */
+  Splice( eDst, eOrg );
+
+  if( ! joiningVertices ) {
+    GLUvertex *newVertex= allocVertex();
+    if (newVertex == NULL) return 0;
+
+    /* We split one vertex into two -- the new vertex is eDst->Org.
+     * Make sure the old vertex points to a valid half-edge.
+     */
+    MakeVertex( newVertex, eDst, eOrg->Org );
+    eOrg->Org->anEdge = eOrg;
+  }
+  if( ! joiningLoops ) {
+    GLUface *newFace= allocFace();  
+    if (newFace == NULL) return 0;
+
+    /* We split one loop into two -- the new loop is eDst->Lface.
+     * Make sure the old face points to a valid half-edge.
+     */
+    MakeFace( newFace, eDst, eOrg->Lface );
+    eOrg->Lface->anEdge = eOrg;
+  }
+
+  return 1;
+}
+
+
+/* __gl_meshDelete( eDel ) removes the edge eDel.  There are several cases:
+ * if (eDel->Lface != eDel->Rface), we join two loops into one; the loop
+ * eDel->Lface is deleted.  Otherwise, we are splitting one loop into two;
+ * the newly created loop will contain eDel->Dst.  If the deletion of eDel
+ * would create isolated vertices, those are deleted as well.
+ *
+ * This function could be implemented as two calls to __gl_meshSplice
+ * plus a few calls to memFree, but this would allocate and delete
+ * unnecessary vertices and faces.
+ */
+int __gl_meshDelete( GLUhalfEdge *eDel )
+{
+  GLUhalfEdge *eDelSym = eDel->Sym;
+  int joiningLoops = FALSE;
+
+  /* First step: disconnect the origin vertex eDel->Org.  We make all
+   * changes to get a consistent mesh in this "intermediate" state.
+   */
+  if( eDel->Lface != eDel->Rface ) {
+    /* We are joining two loops into one -- remove the left face */
+    joiningLoops = TRUE;
+    KillFace( eDel->Lface, eDel->Rface );
+  }
+
+  if( eDel->Onext == eDel ) {
+    KillVertex( eDel->Org, NULL );
+  } else {
+    /* Make sure that eDel->Org and eDel->Rface point to valid half-edges */
+    eDel->Rface->anEdge = eDel->Oprev;
+    eDel->Org->anEdge = eDel->Onext;
+
+    Splice( eDel, eDel->Oprev );
+    if( ! joiningLoops ) {
+      GLUface *newFace= allocFace();
+      if (newFace == NULL) return 0; 
+
+      /* We are splitting one loop into two -- create a new loop for eDel. */
+      MakeFace( newFace, eDel, eDel->Lface );
+    }
+  }
+
+  /* Claim: the mesh is now in a consistent state, except that eDel->Org
+   * may have been deleted.  Now we disconnect eDel->Dst.
+   */
+  if( eDelSym->Onext == eDelSym ) {
+    KillVertex( eDelSym->Org, NULL );
+    KillFace( eDelSym->Lface, NULL );
+  } else {
+    /* Make sure that eDel->Dst and eDel->Lface point to valid half-edges */
+    eDel->Lface->anEdge = eDelSym->Oprev;
+    eDelSym->Org->anEdge = eDelSym->Onext;
+    Splice( eDelSym, eDelSym->Oprev );
+  }
+
+  /* Any isolated vertices or faces have already been freed. */
+  KillEdge( eDel );
+
+  return 1;
+}
+
+
+/******************** Other Edge Operations **********************/
+
+/* All these routines can be implemented with the basic edge
+ * operations above.  They are provided for convenience and efficiency.
+ */
+
+
+/* __gl_meshAddEdgeVertex( eOrg ) creates a new edge eNew such that
+ * eNew == eOrg->Lnext, and eNew->Dst is a newly created vertex.
+ * eOrg and eNew will have the same left face.
+ */
+GLUhalfEdge *__gl_meshAddEdgeVertex( GLUhalfEdge *eOrg )
+{
+  GLUhalfEdge *eNewSym;
+  GLUhalfEdge *eNew = MakeEdge( eOrg );
+  if (eNew == NULL) return NULL;
+
+  eNewSym = eNew->Sym;
+
+  /* Connect the new edge appropriately */
+  Splice( eNew, eOrg->Lnext );
+
+  /* Set the vertex and face information */
+  eNew->Org = eOrg->Dst;
+  {
+    GLUvertex *newVertex= allocVertex();
+    if (newVertex == NULL) return NULL;
+
+    MakeVertex( newVertex, eNewSym, eNew->Org );
+  }
+  eNew->Lface = eNewSym->Lface = eOrg->Lface;
+
+  return eNew;
+}
+
+
+/* __gl_meshSplitEdge( eOrg ) splits eOrg into two edges eOrg and eNew,
+ * such that eNew == eOrg->Lnext.  The new vertex is eOrg->Dst == eNew->Org.
+ * eOrg and eNew will have the same left face.
+ */
+GLUhalfEdge *__gl_meshSplitEdge( GLUhalfEdge *eOrg )
+{
+  GLUhalfEdge *eNew;
+  GLUhalfEdge *tempHalfEdge= __gl_meshAddEdgeVertex( eOrg );
+  if (tempHalfEdge == NULL) return NULL;
+
+  eNew = tempHalfEdge->Sym;
+
+  /* Disconnect eOrg from eOrg->Dst and connect it to eNew->Org */
+  Splice( eOrg->Sym, eOrg->Sym->Oprev );
+  Splice( eOrg->Sym, eNew );
+
+  /* Set the vertex and face information */
+  eOrg->Dst = eNew->Org;
+  eNew->Dst->anEdge = eNew->Sym;	/* may have pointed to eOrg->Sym */
+  eNew->Rface = eOrg->Rface;
+  eNew->winding = eOrg->winding;	/* copy old winding information */
+  eNew->Sym->winding = eOrg->Sym->winding;
+
+  return eNew;
+}
+
+
+/* __gl_meshConnect( eOrg, eDst ) creates a new edge from eOrg->Dst
+ * to eDst->Org, and returns the corresponding half-edge eNew.
+ * If eOrg->Lface == eDst->Lface, this splits one loop into two,
+ * and the newly created loop is eNew->Lface.  Otherwise, two disjoint
+ * loops are merged into one, and the loop eDst->Lface is destroyed.
+ *
+ * If (eOrg == eDst), the new face will have only two edges.
+ * If (eOrg->Lnext == eDst), the old face is reduced to a single edge.
+ * If (eOrg->Lnext->Lnext == eDst), the old face is reduced to two edges.
+ */
+GLUhalfEdge *__gl_meshConnect( GLUhalfEdge *eOrg, GLUhalfEdge *eDst )
+{
+  GLUhalfEdge *eNewSym;
+  int joiningLoops = FALSE;  
+  GLUhalfEdge *eNew = MakeEdge( eOrg );
+  if (eNew == NULL) return NULL;
+
+  eNewSym = eNew->Sym;
+
+  if( eDst->Lface != eOrg->Lface ) {
+    /* We are connecting two disjoint loops -- destroy eDst->Lface */
+    joiningLoops = TRUE;
+    KillFace( eDst->Lface, eOrg->Lface );
+  }
+
+  /* Connect the new edge appropriately */
+  Splice( eNew, eOrg->Lnext );
+  Splice( eNewSym, eDst );
+
+  /* Set the vertex and face information */
+  eNew->Org = eOrg->Dst;
+  eNewSym->Org = eDst->Org;
+  eNew->Lface = eNewSym->Lface = eOrg->Lface;
+
+  /* Make sure the old face points to a valid half-edge */
+  eOrg->Lface->anEdge = eNewSym;
+
+  if( ! joiningLoops ) {
+    GLUface *newFace= allocFace();
+    if (newFace == NULL) return NULL;
+
+    /* We split one loop into two -- the new loop is eNew->Lface */
+    MakeFace( newFace, eNew, eOrg->Lface );
+  }
+  return eNew;
+}
+
+
+/******************** Other Operations **********************/
+
+/* __gl_meshZapFace( fZap ) destroys a face and removes it from the
+ * global face list.  All edges of fZap will have a NULL pointer as their
+ * left face.  Any edges which also have a NULL pointer as their right face
+ * are deleted entirely (along with any isolated vertices this produces).
+ * An entire mesh can be deleted by zapping its faces, one at a time,
+ * in any order.  Zapped faces cannot be used in further mesh operations!
+ */
+void __gl_meshZapFace( GLUface *fZap )
+{
+  GLUhalfEdge *eStart = fZap->anEdge;
+  GLUhalfEdge *e, *eNext, *eSym;
+  GLUface *fPrev, *fNext;
+
+  /* walk around face, deleting edges whose right face is also NULL */
+  eNext = eStart->Lnext;
+  do {
+    e = eNext;
+    eNext = e->Lnext;
+
+    e->Lface = NULL;
+    if( e->Rface == NULL ) {
+      /* delete the edge -- see __gl_MeshDelete above */
+
+      if( e->Onext == e ) {
+	KillVertex( e->Org, NULL );
+      } else {
+	/* Make sure that e->Org points to a valid half-edge */
+	e->Org->anEdge = e->Onext;
+	Splice( e, e->Oprev );
+      }
+      eSym = e->Sym;
+      if( eSym->Onext == eSym ) {
+	KillVertex( eSym->Org, NULL );
+      } else {
+	/* Make sure that eSym->Org points to a valid half-edge */
+	eSym->Org->anEdge = eSym->Onext;
+	Splice( eSym, eSym->Oprev );
+      }
+      KillEdge( e );
+    }
+  } while( e != eStart );
+
+  /* delete from circular doubly-linked list */
+  fPrev = fZap->prev;
+  fNext = fZap->next;
+  fNext->prev = fPrev;
+  fPrev->next = fNext;
+
+  memFree( fZap );
+}
+
+
+/* __gl_meshNewMesh() creates a new mesh with no edges, no vertices,
+ * and no loops (what we usually call a "face").
+ */
+GLUmesh *__gl_meshNewMesh( void )
+{
+  GLUvertex *v;
+  GLUface *f;
+  GLUhalfEdge *e;
+  GLUhalfEdge *eSym;
+  GLUmesh *mesh = (GLUmesh *)memAlloc( sizeof( GLUmesh ));
+  if (mesh == NULL) {
+     return NULL;
+  }
+  
+  v = &mesh->vHead;
+  f = &mesh->fHead;
+  e = &mesh->eHead;
+  eSym = &mesh->eHeadSym;
+
+  v->next = v->prev = v;
+  v->anEdge = NULL;
+  v->data = NULL;
+
+  f->next = f->prev = f;
+  f->anEdge = NULL;
+  f->data = NULL;
+  f->trail = NULL;
+  f->marked = FALSE;
+  f->inside = FALSE;
+
+  e->next = e;
+  e->Sym = eSym;
+  e->Onext = NULL;
+  e->Lnext = NULL;
+  e->Org = NULL;
+  e->Lface = NULL;
+  e->winding = 0;
+  e->activeRegion = NULL;
+
+  eSym->next = eSym;
+  eSym->Sym = e;
+  eSym->Onext = NULL;
+  eSym->Lnext = NULL;
+  eSym->Org = NULL;
+  eSym->Lface = NULL;
+  eSym->winding = 0;
+  eSym->activeRegion = NULL;
+
+  return mesh;
+}
+
+
+/* __gl_meshUnion( mesh1, mesh2 ) forms the union of all structures in
+ * both meshes, and returns the new mesh (the old meshes are destroyed).
+ */
+GLUmesh *__gl_meshUnion( GLUmesh *mesh1, GLUmesh *mesh2 )
+{
+  GLUface *f1 = &mesh1->fHead;
+  GLUvertex *v1 = &mesh1->vHead;
+  GLUhalfEdge *e1 = &mesh1->eHead;
+  GLUface *f2 = &mesh2->fHead;
+  GLUvertex *v2 = &mesh2->vHead;
+  GLUhalfEdge *e2 = &mesh2->eHead;
+
+  /* Add the faces, vertices, and edges of mesh2 to those of mesh1 */
+  if( f2->next != f2 ) {
+    f1->prev->next = f2->next;
+    f2->next->prev = f1->prev;
+    f2->prev->next = f1;
+    f1->prev = f2->prev;
+  }
+
+  if( v2->next != v2 ) {
+    v1->prev->next = v2->next;
+    v2->next->prev = v1->prev;
+    v2->prev->next = v1;
+    v1->prev = v2->prev;
+  }
+
+  if( e2->next != e2 ) {
+    e1->Sym->next->Sym->next = e2->next;
+    e2->next->Sym->next = e1->Sym->next;
+    e2->Sym->next->Sym->next = e1;
+    e1->Sym->next = e2->Sym->next;
+  }
+
+  memFree( mesh2 );
+  return mesh1;
+}
+
+
+#ifdef DELETE_BY_ZAPPING
+
+/* __gl_meshDeleteMesh( mesh ) will free all storage for any valid mesh.
+ */
+void __gl_meshDeleteMesh( GLUmesh *mesh )
+{
+  GLUface *fHead = &mesh->fHead;
+
+  while( fHead->next != fHead ) {
+    __gl_meshZapFace( fHead->next );
+  }
+  assert( mesh->vHead.next == &mesh->vHead );
+
+  memFree( mesh );
+}
+
+#else
+
+/* __gl_meshDeleteMesh( mesh ) will free all storage for any valid mesh.
+ */
+void __gl_meshDeleteMesh( GLUmesh *mesh )
+{
+  GLUface *f, *fNext;
+  GLUvertex *v, *vNext;
+  GLUhalfEdge *e, *eNext;
+
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = fNext ) {
+    fNext = f->next;
+    memFree( f );
+  }
+
+  for( v = mesh->vHead.next; v != &mesh->vHead; v = vNext ) {
+    vNext = v->next;
+    memFree( v );
+  }
+
+  for( e = mesh->eHead.next; e != &mesh->eHead; e = eNext ) {
+    /* One call frees both e and e->Sym (see EdgePair above) */
+    eNext = e->next;
+    memFree( e );
+  }
+
+  memFree( mesh );
+}
+
+#endif
+
+#ifndef NDEBUG
+
+/* __gl_meshCheckMesh( mesh ) checks a mesh for self-consistency.
+ */
+void __gl_meshCheckMesh( GLUmesh *mesh )
+{
+  GLUface *fHead = &mesh->fHead;
+  GLUvertex *vHead = &mesh->vHead;
+  GLUhalfEdge *eHead = &mesh->eHead;
+  GLUface *f, *fPrev;
+  GLUvertex *v, *vPrev;
+  GLUhalfEdge *e, *ePrev;
+
+  fPrev = fHead;
+  for( fPrev = fHead ; (f = fPrev->next) != fHead; fPrev = f) {
+    assert( f->prev == fPrev );
+    e = f->anEdge;
+    do {
+      assert( e->Sym != e );
+      assert( e->Sym->Sym == e );
+      assert( e->Lnext->Onext->Sym == e );
+      assert( e->Onext->Sym->Lnext == e );
+      assert( e->Lface == f );
+      e = e->Lnext;
+    } while( e != f->anEdge );
+  }
+  assert( f->prev == fPrev && f->anEdge == NULL && f->data == NULL );
+
+  vPrev = vHead;
+  for( vPrev = vHead ; (v = vPrev->next) != vHead; vPrev = v) {
+    assert( v->prev == vPrev );
+    e = v->anEdge;
+    do {
+      assert( e->Sym != e );
+      assert( e->Sym->Sym == e );
+      assert( e->Lnext->Onext->Sym == e );
+      assert( e->Onext->Sym->Lnext == e );
+      assert( e->Org == v );
+      e = e->Onext;
+    } while( e != v->anEdge );
+  }
+  assert( v->prev == vPrev && v->anEdge == NULL && v->data == NULL );
+
+  ePrev = eHead;
+  for( ePrev = eHead ; (e = ePrev->next) != eHead; ePrev = e) {
+    assert( e->Sym->next == ePrev->Sym );
+    assert( e->Sym != e );
+    assert( e->Sym->Sym == e );
+    assert( e->Org != NULL );
+    assert( e->Dst != NULL );
+    assert( e->Lnext->Onext->Sym == e );
+    assert( e->Onext->Sym->Lnext == e );
+  }
+  assert( e->Sym->next == ePrev->Sym
+       && e->Sym == &mesh->eHeadSym
+       && e->Sym->Sym == e
+       && e->Org == NULL && e->Dst == NULL
+       && e->Lface == NULL && e->Rface == NULL );
+}
+
+#endif
diff --git a/third_party/glu/libtess/mesh.h b/third_party/glu/libtess/mesh.h
new file mode 100644
index 0000000..da149d7
--- /dev/null
+++ b/third_party/glu/libtess/mesh.h
@@ -0,0 +1,273 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/mesh.h#5 $
+*/
+
+#ifndef __mesh_h_
+#define __mesh_h_
+
+#include <sk_glu.h>
+
+typedef struct GLUmesh GLUmesh; 
+
+typedef struct GLUvertex GLUvertex;
+typedef struct GLUface GLUface;
+typedef struct GLUhalfEdge GLUhalfEdge;
+
+typedef struct ActiveRegion ActiveRegion;	/* Internal data */
+
+/* The mesh structure is similar in spirit, notation, and operations
+ * to the "quad-edge" structure (see L. Guibas and J. Stolfi, Primitives
+ * for the manipulation of general subdivisions and the computation of
+ * Voronoi diagrams, ACM Transactions on Graphics, 4(2):74-123, April 1985).
+ * For a simplified description, see the course notes for CS348a,
+ * "Mathematical Foundations of Computer Graphics", available at the
+ * Stanford bookstore (and taught during the fall quarter).
+ * The implementation also borrows a tiny subset of the graph-based approach
+ * use in Mantyla's Geometric Work Bench (see M. Mantyla, An Introduction
+ * to Sold Modeling, Computer Science Press, Rockville, Maryland, 1988).
+ *
+ * The fundamental data structure is the "half-edge".  Two half-edges
+ * go together to make an edge, but they point in opposite directions.
+ * Each half-edge has a pointer to its mate (the "symmetric" half-edge Sym),
+ * its origin vertex (Org), the face on its left side (Lface), and the
+ * adjacent half-edges in the CCW direction around the origin vertex
+ * (Onext) and around the left face (Lnext).  There is also a "next"
+ * pointer for the global edge list (see below).
+ *
+ * The notation used for mesh navigation:
+ *	Sym   = the mate of a half-edge (same edge, but opposite direction)
+ *	Onext = edge CCW around origin vertex (keep same origin)
+ *	Dnext = edge CCW around destination vertex (keep same dest)
+ *	Lnext = edge CCW around left face (dest becomes new origin)
+ *	Rnext = edge CCW around right face (origin becomes new dest)
+ *
+ * "prev" means to substitute CW for CCW in the definitions above.
+ *
+ * The mesh keeps global lists of all vertices, faces, and edges,
+ * stored as doubly-linked circular lists with a dummy header node.
+ * The mesh stores pointers to these dummy headers (vHead, fHead, eHead).
+ *
+ * The circular edge list is special; since half-edges always occur
+ * in pairs (e and e->Sym), each half-edge stores a pointer in only
+ * one direction.  Starting at eHead and following the e->next pointers
+ * will visit each *edge* once (ie. e or e->Sym, but not both).
+ * e->Sym stores a pointer in the opposite direction, thus it is
+ * always true that e->Sym->next->Sym->next == e.
+ *
+ * Each vertex has a pointer to next and previous vertices in the
+ * circular list, and a pointer to a half-edge with this vertex as
+ * the origin (NULL if this is the dummy header).  There is also a
+ * field "data" for client data.
+ *
+ * Each face has a pointer to the next and previous faces in the
+ * circular list, and a pointer to a half-edge with this face as
+ * the left face (NULL if this is the dummy header).  There is also
+ * a field "data" for client data.
+ *
+ * Note that what we call a "face" is really a loop; faces may consist
+ * of more than one loop (ie. not simply connected), but there is no
+ * record of this in the data structure.  The mesh may consist of
+ * several disconnected regions, so it may not be possible to visit
+ * the entire mesh by starting at a half-edge and traversing the edge
+ * structure.
+ *
+ * The mesh does NOT support isolated vertices; a vertex is deleted along
+ * with its last edge.  Similarly when two faces are merged, one of the
+ * faces is deleted (see __gl_meshDelete below).  For mesh operations,
+ * all face (loop) and vertex pointers must not be NULL.  However, once
+ * mesh manipulation is finished, __gl_MeshZapFace can be used to delete
+ * faces of the mesh, one at a time.  All external faces can be "zapped"
+ * before the mesh is returned to the client; then a NULL face indicates
+ * a region which is not part of the output polygon.
+ */
+
+struct GLUvertex {
+  GLUvertex	*next;		/* next vertex (never NULL) */
+  GLUvertex	*prev;		/* previous vertex (never NULL) */
+  GLUhalfEdge	*anEdge;	/* a half-edge with this origin */
+  void		*data;		/* client's data */
+
+  /* Internal data (keep hidden) */
+  GLdouble	coords[3];	/* vertex location in 3D */
+  GLdouble	s, t;		/* projection onto the sweep plane */
+  long		pqHandle;	/* to allow deletion from priority queue */
+};
+
+struct GLUface {
+  GLUface	*next;		/* next face (never NULL) */
+  GLUface	*prev;		/* previous face (never NULL) */
+  GLUhalfEdge	*anEdge;	/* a half edge with this left face */
+  void		*data;		/* room for client's data */
+
+  /* Internal data (keep hidden) */
+  GLUface	*trail;		/* "stack" for conversion to strips */
+  GLboolean	marked;		/* flag for conversion to strips */
+  GLboolean	inside;		/* this face is in the polygon interior */
+};
+
+struct GLUhalfEdge {
+  GLUhalfEdge	*next;		/* doubly-linked list (prev==Sym->next) */
+  GLUhalfEdge	*Sym;		/* same edge, opposite direction */
+  GLUhalfEdge	*Onext;		/* next edge CCW around origin */
+  GLUhalfEdge	*Lnext;		/* next edge CCW around left face */
+  GLUvertex	*Org;		/* origin vertex (Overtex too long) */
+  GLUface	*Lface;		/* left face */
+
+  /* Internal data (keep hidden) */
+  ActiveRegion	*activeRegion;	/* a region with this upper edge (sweep.c) */
+  int		winding;	/* change in winding number when crossing
+                                   from the right face to the left face */
+};
+
+#define	Rface	Sym->Lface
+#define Dst	Sym->Org
+
+#define Oprev	Sym->Lnext
+#define Lprev   Onext->Sym
+#define Dprev	Lnext->Sym
+#define Rprev	Sym->Onext
+#define Dnext	Rprev->Sym	/* 3 pointers */
+#define Rnext	Oprev->Sym	/* 3 pointers */
+
+
+struct GLUmesh {
+  GLUvertex	vHead;		/* dummy header for vertex list */
+  GLUface	fHead;		/* dummy header for face list */
+  GLUhalfEdge	eHead;		/* dummy header for edge list */
+  GLUhalfEdge	eHeadSym;	/* and its symmetric counterpart */
+};
+
+/* The mesh operations below have three motivations: completeness,
+ * convenience, and efficiency.  The basic mesh operations are MakeEdge,
+ * Splice, and Delete.  All the other edge operations can be implemented
+ * in terms of these.  The other operations are provided for convenience
+ * and/or efficiency.
+ *
+ * When a face is split or a vertex is added, they are inserted into the
+ * global list *before* the existing vertex or face (ie. e->Org or e->Lface).
+ * This makes it easier to process all vertices or faces in the global lists
+ * without worrying about processing the same data twice.  As a convenience,
+ * when a face is split, the "inside" flag is copied from the old face.
+ * Other internal data (v->data, v->activeRegion, f->data, f->marked,
+ * f->trail, e->winding) is set to zero.
+ *
+ * ********************** Basic Edge Operations **************************
+ *
+ * __gl_meshMakeEdge( mesh ) creates one edge, two vertices, and a loop.
+ * The loop (face) consists of the two new half-edges.
+ *
+ * __gl_meshSplice( eOrg, eDst ) is the basic operation for changing the
+ * mesh connectivity and topology.  It changes the mesh so that
+ *	eOrg->Onext <- OLD( eDst->Onext )
+ *	eDst->Onext <- OLD( eOrg->Onext )
+ * where OLD(...) means the value before the meshSplice operation.
+ *
+ * This can have two effects on the vertex structure:
+ *  - if eOrg->Org != eDst->Org, the two vertices are merged together
+ *  - if eOrg->Org == eDst->Org, the origin is split into two vertices
+ * In both cases, eDst->Org is changed and eOrg->Org is untouched.
+ *
+ * Similarly (and independently) for the face structure,
+ *  - if eOrg->Lface == eDst->Lface, one loop is split into two
+ *  - if eOrg->Lface != eDst->Lface, two distinct loops are joined into one
+ * In both cases, eDst->Lface is changed and eOrg->Lface is unaffected.
+ *
+ * __gl_meshDelete( eDel ) removes the edge eDel.  There are several cases:
+ * if (eDel->Lface != eDel->Rface), we join two loops into one; the loop
+ * eDel->Lface is deleted.  Otherwise, we are splitting one loop into two;
+ * the newly created loop will contain eDel->Dst.  If the deletion of eDel
+ * would create isolated vertices, those are deleted as well.
+ *
+ * ********************** Other Edge Operations **************************
+ *
+ * __gl_meshAddEdgeVertex( eOrg ) creates a new edge eNew such that
+ * eNew == eOrg->Lnext, and eNew->Dst is a newly created vertex.
+ * eOrg and eNew will have the same left face.
+ *
+ * __gl_meshSplitEdge( eOrg ) splits eOrg into two edges eOrg and eNew,
+ * such that eNew == eOrg->Lnext.  The new vertex is eOrg->Dst == eNew->Org.
+ * eOrg and eNew will have the same left face.
+ *
+ * __gl_meshConnect( eOrg, eDst ) creates a new edge from eOrg->Dst
+ * to eDst->Org, and returns the corresponding half-edge eNew.
+ * If eOrg->Lface == eDst->Lface, this splits one loop into two,
+ * and the newly created loop is eNew->Lface.  Otherwise, two disjoint
+ * loops are merged into one, and the loop eDst->Lface is destroyed.
+ *
+ * ************************ Other Operations *****************************
+ *
+ * __gl_meshNewMesh() creates a new mesh with no edges, no vertices,
+ * and no loops (what we usually call a "face").
+ *
+ * __gl_meshUnion( mesh1, mesh2 ) forms the union of all structures in
+ * both meshes, and returns the new mesh (the old meshes are destroyed).
+ *
+ * __gl_meshDeleteMesh( mesh ) will free all storage for any valid mesh.
+ *
+ * __gl_meshZapFace( fZap ) destroys a face and removes it from the
+ * global face list.  All edges of fZap will have a NULL pointer as their
+ * left face.  Any edges which also have a NULL pointer as their right face
+ * are deleted entirely (along with any isolated vertices this produces).
+ * An entire mesh can be deleted by zapping its faces, one at a time,
+ * in any order.  Zapped faces cannot be used in further mesh operations!
+ *
+ * __gl_meshCheckMesh( mesh ) checks a mesh for self-consistency.
+ */
+
+GLUhalfEdge	*__gl_meshMakeEdge( GLUmesh *mesh );
+int		__gl_meshSplice( GLUhalfEdge *eOrg, GLUhalfEdge *eDst );
+int		__gl_meshDelete( GLUhalfEdge *eDel );
+
+GLUhalfEdge	*__gl_meshAddEdgeVertex( GLUhalfEdge *eOrg );
+GLUhalfEdge	*__gl_meshSplitEdge( GLUhalfEdge *eOrg );
+GLUhalfEdge	*__gl_meshConnect( GLUhalfEdge *eOrg, GLUhalfEdge *eDst );
+
+GLUmesh		*__gl_meshNewMesh( void );
+GLUmesh		*__gl_meshUnion( GLUmesh *mesh1, GLUmesh *mesh2 );
+void		__gl_meshDeleteMesh( GLUmesh *mesh );
+void		__gl_meshZapFace( GLUface *fZap );
+
+#ifdef NDEBUG
+#define		__gl_meshCheckMesh( mesh )
+#else
+void		__gl_meshCheckMesh( GLUmesh *mesh );
+#endif
+
+#endif
diff --git a/third_party/glu/libtess/normal.c b/third_party/glu/libtess/normal.c
new file mode 100644
index 0000000..8fb22f1
--- /dev/null
+++ b/third_party/glu/libtess/normal.c
@@ -0,0 +1,259 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/normal.c#5 $
+*/
+
+#include "gluos.h"
+#include "mesh.h"
+#include "tess.h"
+#include "normal.h"
+#include <math.h>
+#include <assert.h>
+
+#define TRUE 1
+#define FALSE 0
+
+#define Dot(u,v)	(u[0]*v[0] + u[1]*v[1] + u[2]*v[2])
+
+#if defined(FOR_TRITE_TEST_PROGRAM) || defined(TRUE_PROJECT)
+static void Normalize( GLdouble v[3] )
+{
+  GLdouble len = v[0]*v[0] + v[1]*v[1] + v[2]*v[2];
+
+  assert( len > 0 );
+  len = sqrt( len );
+  v[0] /= len;
+  v[1] /= len;
+  v[2] /= len;
+}
+#endif
+
+#define ABS(x)	((x) < 0 ? -(x) : (x))
+
+static int LongAxis( GLdouble v[3] )
+{
+  int i = 0;
+
+  if( ABS(v[1]) > ABS(v[0]) ) { i = 1; }
+  if( ABS(v[2]) > ABS(v[i]) ) { i = 2; }
+  return i;
+}
+
+static void ComputeNormal( GLUtesselator *tess, GLdouble norm[3] )
+{
+  GLUvertex *v, *v1, *v2;
+  GLdouble c, tLen2, maxLen2;
+  GLdouble maxVal[3], minVal[3], d1[3], d2[3], tNorm[3];
+  GLUvertex *maxVert[3], *minVert[3];
+  GLUvertex *vHead = &tess->mesh->vHead;
+  int i;
+
+  maxVal[0] = maxVal[1] = maxVal[2] = -2 * GLU_TESS_MAX_COORD;
+  minVal[0] = minVal[1] = minVal[2] = 2 * GLU_TESS_MAX_COORD;
+
+  for( v = vHead->next; v != vHead; v = v->next ) {
+    for( i = 0; i < 3; ++i ) {
+      c = v->coords[i];
+      if( c < minVal[i] ) { minVal[i] = c; minVert[i] = v; }
+      if( c > maxVal[i] ) { maxVal[i] = c; maxVert[i] = v; }
+    }
+  }
+
+  /* Find two vertices separated by at least 1/sqrt(3) of the maximum
+   * distance between any two vertices
+   */
+  i = 0;
+  if( maxVal[1] - minVal[1] > maxVal[0] - minVal[0] ) { i = 1; }
+  if( maxVal[2] - minVal[2] > maxVal[i] - minVal[i] ) { i = 2; }
+  if( minVal[i] >= maxVal[i] ) {
+    /* All vertices are the same -- normal doesn't matter */
+    norm[0] = 0; norm[1] = 0; norm[2] = 1;
+    return;
+  }
+
+  /* Look for a third vertex which forms the triangle with maximum area
+   * (Length of normal == twice the triangle area)
+   */
+  maxLen2 = 0;
+  v1 = minVert[i];
+  v2 = maxVert[i];
+  d1[0] = v1->coords[0] - v2->coords[0];
+  d1[1] = v1->coords[1] - v2->coords[1];
+  d1[2] = v1->coords[2] - v2->coords[2];
+  for( v = vHead->next; v != vHead; v = v->next ) {
+    d2[0] = v->coords[0] - v2->coords[0];
+    d2[1] = v->coords[1] - v2->coords[1];
+    d2[2] = v->coords[2] - v2->coords[2];
+    tNorm[0] = d1[1]*d2[2] - d1[2]*d2[1];
+    tNorm[1] = d1[2]*d2[0] - d1[0]*d2[2];
+    tNorm[2] = d1[0]*d2[1] - d1[1]*d2[0];
+    tLen2 = tNorm[0]*tNorm[0] + tNorm[1]*tNorm[1] + tNorm[2]*tNorm[2];
+    if( tLen2 > maxLen2 ) {
+      maxLen2 = tLen2;
+      norm[0] = tNorm[0];
+      norm[1] = tNorm[1];
+      norm[2] = tNorm[2];
+    }
+  }
+
+  if( maxLen2 <= 0 ) {
+    /* All points lie on a single line -- any decent normal will do */
+    norm[0] = norm[1] = norm[2] = 0;
+    norm[LongAxis(d1)] = 1;
+  }
+}
+  
+
+static void CheckOrientation( GLUtesselator *tess )
+{
+  GLdouble area;
+  GLUface *f, *fHead = &tess->mesh->fHead;
+  GLUvertex *v, *vHead = &tess->mesh->vHead;
+  GLUhalfEdge *e;
+
+  /* When we compute the normal automatically, we choose the orientation
+   * so that the the sum of the signed areas of all contours is non-negative.
+   */
+  area = 0;
+  for( f = fHead->next; f != fHead; f = f->next ) {
+    e = f->anEdge;
+    if( e->winding <= 0 ) continue;
+    do {
+      area += (e->Org->s - e->Dst->s) * (e->Org->t + e->Dst->t);
+      e = e->Lnext;
+    } while( e != f->anEdge );
+  }
+  if( area < 0 ) {
+    /* Reverse the orientation by flipping all the t-coordinates */
+    for( v = vHead->next; v != vHead; v = v->next ) {
+      v->t = - v->t;
+    }
+    tess->tUnit[0] = - tess->tUnit[0];
+    tess->tUnit[1] = - tess->tUnit[1];
+    tess->tUnit[2] = - tess->tUnit[2];
+  }
+}
+
+#ifdef FOR_TRITE_TEST_PROGRAM
+#include <stdlib.h>
+extern int RandomSweep;
+#define S_UNIT_X	(RandomSweep ? (2*drand48()-1) : 1.0)
+#define S_UNIT_Y	(RandomSweep ? (2*drand48()-1) : 0.0)
+#else
+#if defined(SLANTED_SWEEP) 
+/* The "feature merging" is not intended to be complete.  There are
+ * special cases where edges are nearly parallel to the sweep line
+ * which are not implemented.  The algorithm should still behave
+ * robustly (ie. produce a reasonable tesselation) in the presence
+ * of such edges, however it may miss features which could have been
+ * merged.  We could minimize this effect by choosing the sweep line
+ * direction to be something unusual (ie. not parallel to one of the
+ * coordinate axes).
+ */
+#define S_UNIT_X	0.50941539564955385	/* Pre-normalized */
+#define S_UNIT_Y	0.86052074622010633
+#else
+#define S_UNIT_X	1.0
+#define S_UNIT_Y	0.0
+#endif
+#endif
+
+/* Determine the polygon normal and project vertices onto the plane
+ * of the polygon.
+ */
+void __gl_projectPolygon( GLUtesselator *tess )
+{
+  GLUvertex *v, *vHead = &tess->mesh->vHead;
+  GLdouble norm[3];
+  GLdouble *sUnit, *tUnit;
+  int i, computedNormal = FALSE;
+
+  norm[0] = tess->normal[0];
+  norm[1] = tess->normal[1];
+  norm[2] = tess->normal[2];
+  if( norm[0] == 0 && norm[1] == 0 && norm[2] == 0 ) {
+    ComputeNormal( tess, norm );
+    computedNormal = TRUE;
+  }
+  sUnit = tess->sUnit;
+  tUnit = tess->tUnit;
+  i = LongAxis( norm );
+
+#if defined(FOR_TRITE_TEST_PROGRAM) || defined(TRUE_PROJECT)
+  /* Choose the initial sUnit vector to be approximately perpendicular
+   * to the normal.
+   */
+  Normalize( norm );
+
+  sUnit[i] = 0;
+  sUnit[(i+1)%3] = S_UNIT_X;
+  sUnit[(i+2)%3] = S_UNIT_Y;
+
+  /* Now make it exactly perpendicular */
+  w = Dot( sUnit, norm );
+  sUnit[0] -= w * norm[0];
+  sUnit[1] -= w * norm[1];
+  sUnit[2] -= w * norm[2];
+  Normalize( sUnit );
+
+  /* Choose tUnit so that (sUnit,tUnit,norm) form a right-handed frame */
+  tUnit[0] = norm[1]*sUnit[2] - norm[2]*sUnit[1];
+  tUnit[1] = norm[2]*sUnit[0] - norm[0]*sUnit[2];
+  tUnit[2] = norm[0]*sUnit[1] - norm[1]*sUnit[0];
+  Normalize( tUnit );
+#else
+  /* Project perpendicular to a coordinate axis -- better numerically */
+  sUnit[i] = 0;
+  sUnit[(i+1)%3] = S_UNIT_X;
+  sUnit[(i+2)%3] = S_UNIT_Y;
+  
+  tUnit[i] = 0;
+  tUnit[(i+1)%3] = (norm[i] > 0) ? -S_UNIT_Y : S_UNIT_Y;
+  tUnit[(i+2)%3] = (norm[i] > 0) ? S_UNIT_X : -S_UNIT_X;
+#endif
+
+  /* Project the vertices onto the sweep plane */
+  for( v = vHead->next; v != vHead; v = v->next ) {
+    v->s = Dot( v->coords, sUnit );
+    v->t = Dot( v->coords, tUnit );
+  }
+  if( computedNormal ) {
+    CheckOrientation( tess );
+  }
+}
diff --git a/third_party/glu/libtess/normal.h b/third_party/glu/libtess/normal.h
new file mode 100644
index 0000000..5a68747
--- /dev/null
+++ b/third_party/glu/libtess/normal.h
@@ -0,0 +1,52 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/normal.h#5 $
+*/
+
+#ifndef __normal_h_
+#define __normal_h_
+
+#include "tess.h"
+
+/* __gl_projectPolygon( tess ) determines the polygon normal
+ * and project vertices onto the plane of the polygon.
+ */
+void __gl_projectPolygon( GLUtesselator *tess );
+
+#endif
diff --git a/third_party/glu/libtess/priorityq-heap.c b/third_party/glu/libtess/priorityq-heap.c
new file mode 100644
index 0000000..163bde6
--- /dev/null
+++ b/third_party/glu/libtess/priorityq-heap.c
@@ -0,0 +1,260 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/priorityq-heap.c#5 $
+*/
+
+#include <stddef.h>
+#include <assert.h>
+#include <limits.h>
+#include "priorityq-heap.h"
+#include "memalloc.h"
+
+#define INIT_SIZE	32
+
+#define TRUE 1
+#define FALSE 0
+
+#ifdef FOR_TRITE_TEST_PROGRAM
+#define LEQ(x,y)	(*pq->leq)(x,y)
+#else
+/* Violates modularity, but a little faster */
+#include "geom.h"
+#define LEQ(x,y)	VertLeq((GLUvertex *)x, (GLUvertex *)y)
+#endif
+
+/* really __gl_pqHeapNewPriorityQ */
+PriorityQ *pqNewPriorityQ( int (*leq)(PQkey key1, PQkey key2) )
+{
+  PriorityQ *pq = (PriorityQ *)memAlloc( sizeof( PriorityQ ));
+  if (pq == NULL) return NULL;
+
+  pq->size = 0;
+  pq->max = INIT_SIZE;
+  pq->nodes = (PQnode *)memAlloc( (INIT_SIZE + 1) * sizeof(pq->nodes[0]) );
+  if (pq->nodes == NULL) {
+     memFree(pq);
+     return NULL;
+  }
+
+  pq->handles = (PQhandleElem *)memAlloc( (INIT_SIZE + 1) * sizeof(pq->handles[0]) );
+  if (pq->handles == NULL) {
+     memFree(pq->nodes);
+     memFree(pq);
+     return NULL;
+  }
+
+  pq->initialized = FALSE;
+  pq->freeList = 0;
+  pq->leq = leq;
+
+  pq->nodes[1].handle = 1;	/* so that Minimum() returns NULL */
+  pq->handles[1].key = NULL;
+  return pq;
+}
+
+/* really __gl_pqHeapDeletePriorityQ */
+void pqDeletePriorityQ( PriorityQ *pq )
+{
+  memFree( pq->handles );
+  memFree( pq->nodes );
+  memFree( pq );
+}
+
+
+static void FloatDown( PriorityQ *pq, long curr )
+{
+  PQnode *n = pq->nodes;
+  PQhandleElem *h = pq->handles;
+  PQhandle hCurr, hChild;
+  long child;
+
+  hCurr = n[curr].handle;
+  for( ;; ) {
+    child = curr << 1;
+    if( child < pq->size && LEQ( h[n[child+1].handle].key,
+				 h[n[child].handle].key )) {
+      ++child;
+    }
+
+    assert(child <= pq->max);
+
+    hChild = n[child].handle;
+    if( child > pq->size || LEQ( h[hCurr].key, h[hChild].key )) {
+      n[curr].handle = hCurr;
+      h[hCurr].node = curr;
+      break;
+    }
+    n[curr].handle = hChild;
+    h[hChild].node = curr;
+    curr = child;
+  }
+}
+
+
+static void FloatUp( PriorityQ *pq, long curr )
+{
+  PQnode *n = pq->nodes;
+  PQhandleElem *h = pq->handles;
+  PQhandle hCurr, hParent;
+  long parent;
+
+  hCurr = n[curr].handle;
+  for( ;; ) {
+    parent = curr >> 1;
+    hParent = n[parent].handle;
+    if( parent == 0 || LEQ( h[hParent].key, h[hCurr].key )) {
+      n[curr].handle = hCurr;
+      h[hCurr].node = curr;
+      break;
+    }
+    n[curr].handle = hParent;
+    h[hParent].node = curr;
+    curr = parent;
+  }
+}
+
+/* really __gl_pqHeapInit */
+void pqInit( PriorityQ *pq )
+{
+  long i;
+
+  /* This method of building a heap is O(n), rather than O(n lg n). */
+
+  for( i = pq->size; i >= 1; --i ) {
+    FloatDown( pq, i );
+  }
+  pq->initialized = TRUE;
+}
+
+/* really __gl_pqHeapInsert */
+/* returns LONG_MAX iff out of memory */
+PQhandle pqInsert( PriorityQ *pq, PQkey keyNew )
+{
+  long curr;
+  PQhandle free;
+
+  curr = ++ pq->size;
+  if( (curr*2) > pq->max ) {
+    PQnode *saveNodes= pq->nodes;
+    PQhandleElem *saveHandles= pq->handles;
+
+    /* If the heap overflows, double its size. */
+    pq->max <<= 1;
+    pq->nodes = (PQnode *)memRealloc( pq->nodes, 
+				     (size_t) 
+				     ((pq->max + 1) * sizeof( pq->nodes[0] )));
+    if (pq->nodes == NULL) {
+       pq->nodes = saveNodes;	/* restore ptr to free upon return */
+       return LONG_MAX;
+    }
+    pq->handles = (PQhandleElem *)memRealloc( pq->handles,
+			                     (size_t)
+			                      ((pq->max + 1) * 
+					       sizeof( pq->handles[0] )));
+    if (pq->handles == NULL) {
+       pq->handles = saveHandles; /* restore ptr to free upon return */
+       return LONG_MAX;
+    }
+  }
+
+  if( pq->freeList == 0 ) {
+    free = curr;
+  } else {
+    free = pq->freeList;
+    pq->freeList = pq->handles[free].node;
+  }
+
+  pq->nodes[curr].handle = free;
+  pq->handles[free].node = curr;
+  pq->handles[free].key = keyNew;
+
+  if( pq->initialized ) {
+    FloatUp( pq, curr );
+  }
+  assert(free != LONG_MAX);
+  return free;
+}
+
+/* really __gl_pqHeapExtractMin */
+PQkey pqExtractMin( PriorityQ *pq )
+{
+  PQnode *n = pq->nodes;
+  PQhandleElem *h = pq->handles;
+  PQhandle hMin = n[1].handle;
+  PQkey min = h[hMin].key;
+
+  if( pq->size > 0 ) {
+    n[1].handle = n[pq->size].handle;
+    h[n[1].handle].node = 1;
+
+    h[hMin].key = NULL;
+    h[hMin].node = pq->freeList;
+    pq->freeList = hMin;
+
+    if( -- pq->size > 0 ) {
+      FloatDown( pq, 1 );
+    }
+  }
+  return min;
+}
+
+/* really __gl_pqHeapDelete */
+void pqDelete( PriorityQ *pq, PQhandle hCurr )
+{
+  PQnode *n = pq->nodes;
+  PQhandleElem *h = pq->handles;
+  long curr;
+
+  assert( hCurr >= 1 && hCurr <= pq->max && h[hCurr].key != NULL );
+
+  curr = h[hCurr].node;
+  n[curr].handle = n[pq->size].handle;
+  h[n[curr].handle].node = curr;
+
+  if( curr <= -- pq->size ) {
+    if( curr <= 1 || LEQ( h[n[curr>>1].handle].key, h[n[curr].handle].key )) {
+      FloatDown( pq, curr );
+    } else {
+      FloatUp( pq, curr );
+    }
+  }
+  h[hCurr].key = NULL;
+  h[hCurr].node = pq->freeList;
+  pq->freeList = hCurr;
+}
diff --git a/third_party/glu/libtess/priorityq-heap.h b/third_party/glu/libtess/priorityq-heap.h
new file mode 100644
index 0000000..02e4434
--- /dev/null
+++ b/third_party/glu/libtess/priorityq-heap.h
@@ -0,0 +1,114 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/priorityq-heap.h#5 $
+*/
+
+#ifndef __priorityq_heap_h_
+#define __priorityq_heap_h_
+
+/* Use #define's so that another heap implementation can use this one */
+
+#define PQkey			PQHeapKey
+#define PQhandle		PQHeapHandle
+#define PriorityQ		PriorityQHeap
+
+#define pqNewPriorityQ(leq)	__gl_pqHeapNewPriorityQ(leq)
+#define pqDeletePriorityQ(pq)	__gl_pqHeapDeletePriorityQ(pq)
+
+/* The basic operations are insertion of a new key (pqInsert),
+ * and examination/extraction of a key whose value is minimum
+ * (pqMinimum/pqExtractMin).  Deletion is also allowed (pqDelete);
+ * for this purpose pqInsert returns a "handle" which is supplied
+ * as the argument.
+ *
+ * An initial heap may be created efficiently by calling pqInsert
+ * repeatedly, then calling pqInit.  In any case pqInit must be called
+ * before any operations other than pqInsert are used.
+ *
+ * If the heap is empty, pqMinimum/pqExtractMin will return a NULL key.
+ * This may also be tested with pqIsEmpty.
+ */
+#define pqInit(pq)		__gl_pqHeapInit(pq)
+#define pqInsert(pq,key)	__gl_pqHeapInsert(pq,key)
+#define pqMinimum(pq)		__gl_pqHeapMinimum(pq)
+#define pqExtractMin(pq)	__gl_pqHeapExtractMin(pq)
+#define pqDelete(pq,handle)	__gl_pqHeapDelete(pq,handle)
+#define pqIsEmpty(pq)		__gl_pqHeapIsEmpty(pq)
+
+
+/* Since we support deletion the data structure is a little more
+ * complicated than an ordinary heap.  "nodes" is the heap itself;
+ * active nodes are stored in the range 1..pq->size.  When the
+ * heap exceeds its allocated size (pq->max), its size doubles.
+ * The children of node i are nodes 2i and 2i+1.
+ *
+ * Each node stores an index into an array "handles".  Each handle
+ * stores a key, plus a pointer back to the node which currently
+ * represents that key (ie. nodes[handles[i].node].handle == i).
+ */
+
+typedef void *PQkey;
+typedef long PQhandle;
+typedef struct PriorityQ PriorityQ;
+
+typedef struct { PQhandle handle; } PQnode;
+typedef struct { PQkey key; PQhandle node; } PQhandleElem;
+
+struct PriorityQ {
+  PQnode	*nodes;
+  PQhandleElem	*handles;
+  long		size, max;
+  PQhandle	freeList;
+  int		initialized;
+  int		(*leq)(PQkey key1, PQkey key2);
+};
+  
+PriorityQ	*pqNewPriorityQ( int (*leq)(PQkey key1, PQkey key2) );
+void		pqDeletePriorityQ( PriorityQ *pq );
+
+void		pqInit( PriorityQ *pq );
+PQhandle	pqInsert( PriorityQ *pq, PQkey key );
+PQkey		pqExtractMin( PriorityQ *pq );
+void		pqDelete( PriorityQ *pq, PQhandle handle );
+
+
+#define __gl_pqHeapMinimum(pq)	((pq)->handles[(pq)->nodes[1].handle].key)
+#define __gl_pqHeapIsEmpty(pq)	((pq)->size == 0)
+
+#endif
diff --git a/third_party/glu/libtess/priorityq-sort.h b/third_party/glu/libtess/priorityq-sort.h
new file mode 100644
index 0000000..12c2f0d
--- /dev/null
+++ b/third_party/glu/libtess/priorityq-sort.h
@@ -0,0 +1,124 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/priorityq-sort.h#5 $
+*/
+
+#ifndef __priorityq_sort_h_
+#define __priorityq_sort_h_
+
+#include "priorityq-heap.h"
+
+#undef PQkey
+#undef PQhandle
+#undef PriorityQ
+#undef pqNewPriorityQ
+#undef pqDeletePriorityQ
+#undef pqInit
+#undef pqInsert
+#undef pqMinimum
+#undef pqExtractMin
+#undef pqDelete
+#undef pqIsEmpty
+
+/* Use #define's so that another heap implementation can use this one */
+
+#define PQkey			PQSortKey
+#define PQhandle		PQSortHandle
+#define PriorityQ		PriorityQSort
+
+#define pqNewPriorityQ(leq)	__gl_pqSortNewPriorityQ(leq)
+#define pqDeletePriorityQ(pq)	__gl_pqSortDeletePriorityQ(pq)
+
+/* The basic operations are insertion of a new key (pqInsert),
+ * and examination/extraction of a key whose value is minimum
+ * (pqMinimum/pqExtractMin).  Deletion is also allowed (pqDelete);
+ * for this purpose pqInsert returns a "handle" which is supplied
+ * as the argument.
+ *
+ * An initial heap may be created efficiently by calling pqInsert
+ * repeatedly, then calling pqInit.  In any case pqInit must be called
+ * before any operations other than pqInsert are used.
+ *
+ * If the heap is empty, pqMinimum/pqExtractMin will return a NULL key.
+ * This may also be tested with pqIsEmpty.
+ */
+#define pqInit(pq)		__gl_pqSortInit(pq)
+#define pqInsert(pq,key)	__gl_pqSortInsert(pq,key)
+#define pqMinimum(pq)		__gl_pqSortMinimum(pq)
+#define pqExtractMin(pq)	__gl_pqSortExtractMin(pq)
+#define pqDelete(pq,handle)	__gl_pqSortDelete(pq,handle)
+#define pqIsEmpty(pq)		__gl_pqSortIsEmpty(pq)
+
+
+/* Since we support deletion the data structure is a little more
+ * complicated than an ordinary heap.  "nodes" is the heap itself;
+ * active nodes are stored in the range 1..pq->size.  When the
+ * heap exceeds its allocated size (pq->max), its size doubles.
+ * The children of node i are nodes 2i and 2i+1.
+ *
+ * Each node stores an index into an array "handles".  Each handle
+ * stores a key, plus a pointer back to the node which currently
+ * represents that key (ie. nodes[handles[i].node].handle == i).
+ */
+
+typedef PQHeapKey PQkey;
+typedef PQHeapHandle PQhandle;
+typedef struct PriorityQ PriorityQ;
+
+struct PriorityQ {
+  PriorityQHeap	*heap;
+  PQkey		*keys;
+  PQkey		**order;
+  PQhandle	size, max;
+  int		initialized;
+  int		(*leq)(PQkey key1, PQkey key2);
+};
+  
+PriorityQ	*pqNewPriorityQ( int (*leq)(PQkey key1, PQkey key2) );
+void		pqDeletePriorityQ( PriorityQ *pq );
+
+int		pqInit( PriorityQ *pq );
+PQhandle	pqInsert( PriorityQ *pq, PQkey key );
+PQkey		pqExtractMin( PriorityQ *pq );
+void		pqDelete( PriorityQ *pq, PQhandle handle );
+
+PQkey		pqMinimum( PriorityQ *pq );
+int		pqIsEmpty( PriorityQ *pq );
+
+#endif
diff --git a/third_party/glu/libtess/priorityq.c b/third_party/glu/libtess/priorityq.c
new file mode 100644
index 0000000..4b9e1ae
--- /dev/null
+++ b/third_party/glu/libtess/priorityq.c
@@ -0,0 +1,267 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/priorityq.c#5 $
+*/
+
+#include "gluos.h"
+#include <stddef.h>
+#include <assert.h>
+#include <limits.h>		/* LONG_MAX */
+#include "memalloc.h"
+
+/* Include all the code for the regular heap-based queue here. */
+
+#include "priorityq-heap.c"
+
+/* Now redefine all the function names to map to their "Sort" versions. */
+
+#include "priorityq-sort.h"
+
+/* really __gl_pqSortNewPriorityQ */
+PriorityQ *pqNewPriorityQ( int (*leq)(PQkey key1, PQkey key2) )
+{
+  PriorityQ *pq = (PriorityQ *)memAlloc( sizeof( PriorityQ ));
+  if (pq == NULL) return NULL;
+
+  pq->heap = __gl_pqHeapNewPriorityQ( leq );
+  if (pq->heap == NULL) {
+     memFree(pq);
+     return NULL;
+  }
+
+  pq->keys = (PQHeapKey *)memAlloc( INIT_SIZE * sizeof(pq->keys[0]) );
+  if (pq->keys == NULL) {
+     __gl_pqHeapDeletePriorityQ(pq->heap);
+     memFree(pq);
+     return NULL;
+  }
+
+  pq->size = 0;
+  pq->max = INIT_SIZE;
+  pq->initialized = FALSE;
+  pq->leq = leq;
+  return pq;
+}
+
+/* really __gl_pqSortDeletePriorityQ */
+void pqDeletePriorityQ( PriorityQ *pq )
+{
+  assert(pq != NULL); 
+  if (pq->heap != NULL) __gl_pqHeapDeletePriorityQ( pq->heap );
+  if (pq->order != NULL) memFree( pq->order );
+  if (pq->keys != NULL) memFree( pq->keys );
+  memFree( pq );
+}
+
+
+#define LT(x,y)		(! LEQ(y,x))
+#define GT(x,y)		(! LEQ(x,y))
+#define Swap(a,b)	do{PQkey *tmp = *a; *a = *b; *b = tmp;}while(0)
+
+/* really __gl_pqSortInit */
+int pqInit( PriorityQ *pq )
+{
+  PQkey **p, **r, **i, **j, *piv;
+  struct { PQkey **p, **r; } Stack[50], *top = Stack;
+  unsigned long seed = 2016473283;
+
+  /* Create an array of indirect pointers to the keys, so that we
+   * the handles we have returned are still valid.
+   */
+/*
+  pq->order = (PQHeapKey **)memAlloc( (size_t)
+                                  (pq->size * sizeof(pq->order[0])) );
+*/
+  pq->order = (PQHeapKey **)memAlloc( (size_t)
+                                  ((pq->size+1) * sizeof(pq->order[0])) );
+/* the previous line is a patch to compensate for the fact that IBM */
+/* machines return a null on a malloc of zero bytes (unlike SGI),   */
+/* so we have to put in this defense to guard against a memory      */
+/* fault four lines down. from fossum@austin.ibm.com.               */
+  if (pq->order == NULL) return 0;
+
+  p = pq->order;
+  r = p + pq->size - 1;
+  for( piv = pq->keys, i = p; i <= r; ++piv, ++i ) {
+    *i = piv;
+  }
+
+  /* Sort the indirect pointers in descending order,
+   * using randomized Quicksort
+   */
+  top->p = p; top->r = r; ++top;
+  while( --top >= Stack ) {
+    p = top->p;
+    r = top->r;
+    while( r > p + 10 ) {
+      seed = seed * 1539415821 + 1;
+      i = p + seed % (r - p + 1);
+      piv = *i;
+      *i = *p;
+      *p = piv;
+      i = p - 1;
+      j = r + 1;
+      do {
+	do { ++i; } while( GT( **i, *piv ));
+	do { --j; } while( LT( **j, *piv ));
+	Swap( i, j );
+      } while( i < j );
+      Swap( i, j );	/* Undo last swap */
+      if( i - p < r - j ) {
+	top->p = j+1; top->r = r; ++top;
+	r = i-1;
+      } else {
+	top->p = p; top->r = i-1; ++top;
+	p = j+1;
+      }
+    }
+    /* Insertion sort small lists */
+    for( i = p+1; i <= r; ++i ) {
+      piv = *i;
+      for( j = i; j > p && LT( **(j-1), *piv ); --j ) {
+	*j = *(j-1);
+      }
+      *j = piv;
+    }
+  }
+  pq->max = pq->size;
+  pq->initialized = TRUE;
+  __gl_pqHeapInit( pq->heap );	/* always succeeds */
+
+#ifndef NDEBUG
+  p = pq->order;
+  r = p + pq->size - 1;
+  for( i = p; i < r; ++i ) {
+    assert( LEQ( **(i+1), **i ));
+  }
+#endif
+
+  return 1;
+}
+
+/* really __gl_pqSortInsert */
+/* returns LONG_MAX iff out of memory */ 
+PQhandle pqInsert( PriorityQ *pq, PQkey keyNew )
+{
+  long curr;
+
+  if( pq->initialized ) {
+    return __gl_pqHeapInsert( pq->heap, keyNew );
+  }
+  curr = pq->size;
+  if( ++ pq->size >= pq->max ) {
+    PQkey *saveKey= pq->keys;
+
+    /* If the heap overflows, double its size. */
+    pq->max <<= 1;
+    pq->keys = (PQHeapKey *)memRealloc( pq->keys, 
+	 	                        (size_t)
+	                                 (pq->max * sizeof( pq->keys[0] )));
+    if (pq->keys == NULL) {	
+       pq->keys = saveKey;	/* restore ptr to free upon return */
+       return LONG_MAX;
+    }
+  }
+  assert(curr != LONG_MAX);	
+  pq->keys[curr] = keyNew;
+
+  /* Negative handles index the sorted array. */
+  return -(curr+1);
+}
+
+/* really __gl_pqSortExtractMin */
+PQkey pqExtractMin( PriorityQ *pq )
+{
+  PQkey sortMin, heapMin;
+
+  if( pq->size == 0 ) {
+    return __gl_pqHeapExtractMin( pq->heap );
+  }
+  sortMin = *(pq->order[pq->size-1]);
+  if( ! __gl_pqHeapIsEmpty( pq->heap )) {
+    heapMin = __gl_pqHeapMinimum( pq->heap );
+    if( LEQ( heapMin, sortMin )) {
+      return __gl_pqHeapExtractMin( pq->heap );
+    }
+  }
+  do {
+    -- pq->size;
+  } while( pq->size > 0 && *(pq->order[pq->size-1]) == NULL );
+  return sortMin;
+}
+
+/* really __gl_pqSortMinimum */
+PQkey pqMinimum( PriorityQ *pq )
+{
+  PQkey sortMin, heapMin;
+
+  if( pq->size == 0 ) {
+    return __gl_pqHeapMinimum( pq->heap );
+  }
+  sortMin = *(pq->order[pq->size-1]);
+  if( ! __gl_pqHeapIsEmpty( pq->heap )) {
+    heapMin = __gl_pqHeapMinimum( pq->heap );
+    if( LEQ( heapMin, sortMin )) {
+      return heapMin;
+    }
+  }
+  return sortMin;
+}
+
+/* really __gl_pqSortIsEmpty */
+int pqIsEmpty( PriorityQ *pq )
+{
+  return (pq->size == 0) && __gl_pqHeapIsEmpty( pq->heap );
+}
+
+/* really __gl_pqSortDelete */
+void pqDelete( PriorityQ *pq, PQhandle curr )
+{
+  if( curr >= 0 ) {
+    __gl_pqHeapDelete( pq->heap, curr );
+    return;
+  }
+  curr = -(curr+1);
+  assert( curr < pq->max && pq->keys[curr] != NULL );
+
+  pq->keys[curr] = NULL;
+  while( pq->size > 0 && *(pq->order[pq->size-1]) == NULL ) {
+    -- pq->size;
+  }
+}
diff --git a/third_party/glu/libtess/priorityq.h b/third_party/glu/libtess/priorityq.h
new file mode 100644
index 0000000..3ae6b0c
--- /dev/null
+++ b/third_party/glu/libtess/priorityq.h
@@ -0,0 +1,124 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/priorityq.h#5 $
+*/
+
+#ifndef __priorityq_sort_h_
+#define __priorityq_sort_h_
+
+#include "priorityq-heap.h"
+
+#undef PQkey
+#undef PQhandle
+#undef PriorityQ
+#undef pqNewPriorityQ
+#undef pqDeletePriorityQ
+#undef pqInit
+#undef pqInsert
+#undef pqMinimum
+#undef pqExtractMin
+#undef pqDelete
+#undef pqIsEmpty
+
+/* Use #define's so that another heap implementation can use this one */
+
+#define PQkey			PQSortKey
+#define PQhandle		PQSortHandle
+#define PriorityQ		PriorityQSort
+
+#define pqNewPriorityQ(leq)	__gl_pqSortNewPriorityQ(leq)
+#define pqDeletePriorityQ(pq)	__gl_pqSortDeletePriorityQ(pq)
+
+/* The basic operations are insertion of a new key (pqInsert),
+ * and examination/extraction of a key whose value is minimum
+ * (pqMinimum/pqExtractMin).  Deletion is also allowed (pqDelete);
+ * for this purpose pqInsert returns a "handle" which is supplied
+ * as the argument.
+ *
+ * An initial heap may be created efficiently by calling pqInsert
+ * repeatedly, then calling pqInit.  In any case pqInit must be called
+ * before any operations other than pqInsert are used.
+ *
+ * If the heap is empty, pqMinimum/pqExtractMin will return a NULL key.
+ * This may also be tested with pqIsEmpty.
+ */
+#define pqInit(pq)		__gl_pqSortInit(pq)
+#define pqInsert(pq,key)	__gl_pqSortInsert(pq,key)
+#define pqMinimum(pq)		__gl_pqSortMinimum(pq)
+#define pqExtractMin(pq)	__gl_pqSortExtractMin(pq)
+#define pqDelete(pq,handle)	__gl_pqSortDelete(pq,handle)
+#define pqIsEmpty(pq)		__gl_pqSortIsEmpty(pq)
+
+
+/* Since we support deletion the data structure is a little more
+ * complicated than an ordinary heap.  "nodes" is the heap itself;
+ * active nodes are stored in the range 1..pq->size.  When the
+ * heap exceeds its allocated size (pq->max), its size doubles.
+ * The children of node i are nodes 2i and 2i+1.
+ *
+ * Each node stores an index into an array "handles".  Each handle
+ * stores a key, plus a pointer back to the node which currently
+ * represents that key (ie. nodes[handles[i].node].handle == i).
+ */
+
+typedef PQHeapKey PQkey;
+typedef PQHeapHandle PQhandle;
+typedef struct PriorityQ PriorityQ;
+
+struct PriorityQ {
+  PriorityQHeap	*heap;
+  PQkey		*keys;
+  PQkey		**order;
+  PQhandle	size, max;
+  int		initialized;
+  int		(*leq)(PQkey key1, PQkey key2);
+};
+  
+PriorityQ	*pqNewPriorityQ( int (*leq)(PQkey key1, PQkey key2) );
+void		pqDeletePriorityQ( PriorityQ *pq );
+
+int		pqInit( PriorityQ *pq );
+PQhandle	pqInsert( PriorityQ *pq, PQkey key );
+PQkey		pqExtractMin( PriorityQ *pq );
+void		pqDelete( PriorityQ *pq, PQhandle handle );
+
+PQkey		pqMinimum( PriorityQ *pq );
+int		pqIsEmpty( PriorityQ *pq );
+
+#endif
diff --git a/third_party/glu/libtess/render.c b/third_party/glu/libtess/render.c
new file mode 100644
index 0000000..aff1298
--- /dev/null
+++ b/third_party/glu/libtess/render.c
@@ -0,0 +1,505 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/render.c#5 $
+*/
+
+#include "gluos.h"
+#include <assert.h>
+#include <stddef.h>
+#include "mesh.h"
+#include "tess.h"
+#include "render.h"
+
+#define TRUE 1
+#define FALSE 0
+
+/* This structure remembers the information we need about a primitive
+ * to be able to render it later, once we have determined which
+ * primitive is able to use the most triangles.
+ */
+struct FaceCount {
+  long		size;		/* number of triangles used */
+  GLUhalfEdge	*eStart;	/* edge where this primitive starts */
+  void		(*render)(GLUtesselator *, GLUhalfEdge *, long);
+                                /* routine to render this primitive */
+};
+
+static struct FaceCount MaximumFan( GLUhalfEdge *eOrig );
+static struct FaceCount MaximumStrip( GLUhalfEdge *eOrig );
+
+static void RenderFan( GLUtesselator *tess, GLUhalfEdge *eStart, long size );
+static void RenderStrip( GLUtesselator *tess, GLUhalfEdge *eStart, long size );
+static void RenderTriangle( GLUtesselator *tess, GLUhalfEdge *eStart,
+			    long size );
+
+static void RenderMaximumFaceGroup( GLUtesselator *tess, GLUface *fOrig );
+static void RenderLonelyTriangles( GLUtesselator *tess, GLUface *head );
+
+
+
+/************************ Strips and Fans decomposition ******************/
+
+/* __gl_renderMesh( tess, mesh ) takes a mesh and breaks it into triangle
+ * fans, strips, and separate triangles.  A substantial effort is made
+ * to use as few rendering primitives as possible (ie. to make the fans
+ * and strips as large as possible).
+ *
+ * The rendering output is provided as callbacks (see the api).
+ */
+void __gl_renderMesh( GLUtesselator *tess, GLUmesh *mesh )
+{
+  GLUface *f;
+
+  /* Make a list of separate triangles so we can render them all at once */
+  tess->lonelyTriList = NULL;
+
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = f->next ) {
+    f->marked = FALSE;
+  }
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = f->next ) {
+
+    /* We examine all faces in an arbitrary order.  Whenever we find
+     * an unprocessed face F, we output a group of faces including F
+     * whose size is maximum.
+     */
+    if( f->inside && ! f->marked ) {
+      RenderMaximumFaceGroup( tess, f );
+      assert( f->marked );
+    }
+  }
+  if( tess->lonelyTriList != NULL ) {
+    RenderLonelyTriangles( tess, tess->lonelyTriList );
+    tess->lonelyTriList = NULL;
+  }
+}
+
+
+static void RenderMaximumFaceGroup( GLUtesselator *tess, GLUface *fOrig )
+{
+  /* We want to find the largest triangle fan or strip of unmarked faces
+   * which includes the given face fOrig.  There are 3 possible fans
+   * passing through fOrig (one centered at each vertex), and 3 possible
+   * strips (one for each CCW permutation of the vertices).  Our strategy
+   * is to try all of these, and take the primitive which uses the most
+   * triangles (a greedy approach).
+   */
+  GLUhalfEdge *e = fOrig->anEdge;
+  struct FaceCount max, newFace;
+
+  max.size = 1;
+  max.eStart = e;
+  max.render = &RenderTriangle;
+
+  if( ! tess->flagBoundary ) {
+    newFace = MaximumFan( e ); if( newFace.size > max.size ) { max = newFace; }
+    newFace = MaximumFan( e->Lnext ); if( newFace.size > max.size ) { max = newFace; }
+    newFace = MaximumFan( e->Lprev ); if( newFace.size > max.size ) { max = newFace; }
+
+    newFace = MaximumStrip( e ); if( newFace.size > max.size ) { max = newFace; }
+    newFace = MaximumStrip( e->Lnext ); if( newFace.size > max.size ) { max = newFace; }
+    newFace = MaximumStrip( e->Lprev ); if( newFace.size > max.size ) { max = newFace; }
+  }
+  (*(max.render))( tess, max.eStart, max.size );
+}
+
+
+/* Macros which keep track of faces we have marked temporarily, and allow
+ * us to backtrack when necessary.  With triangle fans, this is not
+ * really necessary, since the only awkward case is a loop of triangles
+ * around a single origin vertex.  However with strips the situation is
+ * more complicated, and we need a general tracking method like the
+ * one here.
+ */
+#define Marked(f)	(! (f)->inside || (f)->marked)
+
+#define AddToTrail(f,t)	((f)->trail = (t), (t) = (f), (f)->marked = TRUE)
+
+#define FreeTrail(t)	do { \
+			  while( (t) != NULL ) { \
+			    (t)->marked = FALSE; t = (t)->trail; \
+			  } \
+			} while(0) /* absorb trailing semicolon */
+
+
+
+static struct FaceCount MaximumFan( GLUhalfEdge *eOrig )
+{
+  /* eOrig->Lface is the face we want to render.  We want to find the size
+   * of a maximal fan around eOrig->Org.  To do this we just walk around
+   * the origin vertex as far as possible in both directions.
+   */
+  struct FaceCount newFace = { 0, NULL, &RenderFan };
+  GLUface *trail = NULL;
+  GLUhalfEdge *e;
+
+  for( e = eOrig; ! Marked( e->Lface ); e = e->Onext ) {
+    AddToTrail( e->Lface, trail );
+    ++newFace.size;
+  }
+  for( e = eOrig; ! Marked( e->Rface ); e = e->Oprev ) {
+    AddToTrail( e->Rface, trail );
+    ++newFace.size;
+  }
+  newFace.eStart = e;
+  /*LINTED*/
+  FreeTrail( trail );
+  return newFace;
+}
+
+
+#define IsEven(n)	(((n) & 1) == 0)
+
+static struct FaceCount MaximumStrip( GLUhalfEdge *eOrig )
+{
+  /* Here we are looking for a maximal strip that contains the vertices
+   * eOrig->Org, eOrig->Dst, eOrig->Lnext->Dst (in that order or the
+   * reverse, such that all triangles are oriented CCW).
+   *
+   * Again we walk forward and backward as far as possible.  However for
+   * strips there is a twist: to get CCW orientations, there must be
+   * an *even* number of triangles in the strip on one side of eOrig.
+   * We walk the strip starting on a side with an even number of triangles;
+   * if both side have an odd number, we are forced to shorten one side.
+   */
+  struct FaceCount newFace = { 0, NULL, &RenderStrip };
+  long headSize = 0, tailSize = 0;
+  GLUface *trail = NULL;
+  GLUhalfEdge *e, *eTail, *eHead;
+
+  for( e = eOrig; ! Marked( e->Lface ); ++tailSize, e = e->Onext ) {
+    AddToTrail( e->Lface, trail );
+    ++tailSize;
+    e = e->Dprev;
+    if( Marked( e->Lface )) break;
+    AddToTrail( e->Lface, trail );
+  }
+  eTail = e;
+
+  for( e = eOrig; ! Marked( e->Rface ); ++headSize, e = e->Dnext ) {
+    AddToTrail( e->Rface, trail );
+    ++headSize;
+    e = e->Oprev;
+    if( Marked( e->Rface )) break;
+    AddToTrail( e->Rface, trail );
+  }
+  eHead = e;
+
+  newFace.size = tailSize + headSize;
+  if( IsEven( tailSize )) {
+    newFace.eStart = eTail->Sym;
+  } else if( IsEven( headSize )) {
+    newFace.eStart = eHead;
+  } else {
+    /* Both sides have odd length, we must shorten one of them.  In fact,
+     * we must start from eHead to guarantee inclusion of eOrig->Lface.
+     */
+    --newFace.size;
+    newFace.eStart = eHead->Onext;
+  }
+  /*LINTED*/
+  FreeTrail( trail );
+  return newFace;
+}
+
+
+static void RenderTriangle( GLUtesselator *tess, GLUhalfEdge *e, long size )
+{
+  /* Just add the triangle to a triangle list, so we can render all
+   * the separate triangles at once.
+   */
+  assert( size == 1 );
+  AddToTrail( e->Lface, tess->lonelyTriList );
+}
+
+
+static void RenderLonelyTriangles( GLUtesselator *tess, GLUface *f )
+{
+  /* Now we render all the separate triangles which could not be
+   * grouped into a triangle fan or strip.
+   */
+  GLUhalfEdge *e;
+  int newState;
+  int edgeState = -1;	/* force edge state output for first vertex */
+
+  CALL_BEGIN_OR_BEGIN_DATA( GL_TRIANGLES );
+
+  for( ; f != NULL; f = f->trail ) {
+    /* Loop once for each edge (there will always be 3 edges) */
+
+    e = f->anEdge;
+    do {
+      if( tess->flagBoundary ) {
+	/* Set the "edge state" to TRUE just before we output the
+	 * first vertex of each edge on the polygon boundary.
+	 */
+	newState = ! e->Rface->inside;
+	if( edgeState != newState ) {
+	  edgeState = newState;
+          CALL_EDGE_FLAG_OR_EDGE_FLAG_DATA( edgeState );
+	}
+      }
+      CALL_VERTEX_OR_VERTEX_DATA( e->Org->data );
+
+      e = e->Lnext;
+    } while( e != f->anEdge );
+  }
+  CALL_END_OR_END_DATA();
+}
+
+
+static void RenderFan( GLUtesselator *tess, GLUhalfEdge *e, long size )
+{
+  /* Render as many CCW triangles as possible in a fan starting from
+   * edge "e".  The fan *should* contain exactly "size" triangles
+   * (otherwise we've goofed up somewhere).
+   */
+  CALL_BEGIN_OR_BEGIN_DATA( GL_TRIANGLE_FAN ); 
+  CALL_VERTEX_OR_VERTEX_DATA( e->Org->data ); 
+  CALL_VERTEX_OR_VERTEX_DATA( e->Dst->data ); 
+
+  while( ! Marked( e->Lface )) {
+    e->Lface->marked = TRUE;
+    --size;
+    e = e->Onext;
+    CALL_VERTEX_OR_VERTEX_DATA( e->Dst->data ); 
+  }
+
+  assert( size == 0 );
+  CALL_END_OR_END_DATA();
+}
+
+
+static void RenderStrip( GLUtesselator *tess, GLUhalfEdge *e, long size )
+{
+  /* Render as many CCW triangles as possible in a strip starting from
+   * edge "e".  The strip *should* contain exactly "size" triangles
+   * (otherwise we've goofed up somewhere).
+   */
+  CALL_BEGIN_OR_BEGIN_DATA( GL_TRIANGLE_STRIP );
+  CALL_VERTEX_OR_VERTEX_DATA( e->Org->data ); 
+  CALL_VERTEX_OR_VERTEX_DATA( e->Dst->data ); 
+
+  while( ! Marked( e->Lface )) {
+    e->Lface->marked = TRUE;
+    --size;
+    e = e->Dprev;
+    CALL_VERTEX_OR_VERTEX_DATA( e->Org->data ); 
+    if( Marked( e->Lface )) break;
+
+    e->Lface->marked = TRUE;
+    --size;
+    e = e->Onext;
+    CALL_VERTEX_OR_VERTEX_DATA( e->Dst->data ); 
+  }
+
+  assert( size == 0 );
+  CALL_END_OR_END_DATA();
+}
+
+
+/************************ Boundary contour decomposition ******************/
+
+/* __gl_renderBoundary( tess, mesh ) takes a mesh, and outputs one
+ * contour for each face marked "inside".  The rendering output is
+ * provided as callbacks (see the api).
+ */
+void __gl_renderBoundary( GLUtesselator *tess, GLUmesh *mesh )
+{
+  GLUface *f;
+  GLUhalfEdge *e;
+
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = f->next ) {
+    if( f->inside ) {
+      CALL_BEGIN_OR_BEGIN_DATA( GL_LINE_LOOP );
+      e = f->anEdge;
+      do {
+        CALL_VERTEX_OR_VERTEX_DATA( e->Org->data ); 
+	e = e->Lnext;
+      } while( e != f->anEdge );
+      CALL_END_OR_END_DATA();
+    }
+  }
+}
+
+
+/************************ Quick-and-dirty decomposition ******************/
+
+#define SIGN_INCONSISTENT 2
+
+static int ComputeNormal( GLUtesselator *tess, GLdouble norm[3], int check )
+/*
+ * If check==FALSE, we compute the polygon normal and place it in norm[].
+ * If check==TRUE, we check that each triangle in the fan from v0 has a
+ * consistent orientation with respect to norm[].  If triangles are
+ * consistently oriented CCW, return 1; if CW, return -1; if all triangles
+ * are degenerate return 0; otherwise (no consistent orientation) return
+ * SIGN_INCONSISTENT.
+ */
+{
+  CachedVertex *v0 = tess->cache;
+  CachedVertex *vn = v0 + tess->cacheCount;
+  CachedVertex *vc;
+  GLdouble dot, xc, yc, zc, xp, yp, zp, n[3];
+  int sign = 0;
+
+  /* Find the polygon normal.  It is important to get a reasonable
+   * normal even when the polygon is self-intersecting (eg. a bowtie).
+   * Otherwise, the computed normal could be very tiny, but perpendicular
+   * to the true plane of the polygon due to numerical noise.  Then all
+   * the triangles would appear to be degenerate and we would incorrectly
+   * decompose the polygon as a fan (or simply not render it at all).
+   *
+   * We use a sum-of-triangles normal algorithm rather than the more
+   * efficient sum-of-trapezoids method (used in CheckOrientation()
+   * in normal.c).  This lets us explicitly reverse the signed area
+   * of some triangles to get a reasonable normal in the self-intersecting
+   * case.
+   */
+  if( ! check ) {
+    norm[0] = norm[1] = norm[2] = 0.0;
+  }
+
+  vc = v0 + 1;
+  xc = vc->coords[0] - v0->coords[0];
+  yc = vc->coords[1] - v0->coords[1];
+  zc = vc->coords[2] - v0->coords[2];
+  while( ++vc < vn ) {
+    xp = xc; yp = yc; zp = zc;
+    xc = vc->coords[0] - v0->coords[0];
+    yc = vc->coords[1] - v0->coords[1];
+    zc = vc->coords[2] - v0->coords[2];
+
+    /* Compute (vp - v0) cross (vc - v0) */
+    n[0] = yp*zc - zp*yc;
+    n[1] = zp*xc - xp*zc;
+    n[2] = xp*yc - yp*xc;
+
+    dot = n[0]*norm[0] + n[1]*norm[1] + n[2]*norm[2];
+    if( ! check ) {
+      /* Reverse the contribution of back-facing triangles to get
+       * a reasonable normal for self-intersecting polygons (see above)
+       */
+      if( dot >= 0 ) {
+	norm[0] += n[0]; norm[1] += n[1]; norm[2] += n[2];
+      } else {
+	norm[0] -= n[0]; norm[1] -= n[1]; norm[2] -= n[2];
+      }
+    } else if( dot != 0 ) {
+      /* Check the new orientation for consistency with previous triangles */
+      if( dot > 0 ) {
+	if( sign < 0 ) return SIGN_INCONSISTENT;
+	sign = 1;
+      } else {
+	if( sign > 0 ) return SIGN_INCONSISTENT;
+	sign = -1;
+      }
+    }
+  }
+  return sign;
+}
+
+/* __gl_renderCache( tess ) takes a single contour and tries to render it
+ * as a triangle fan.  This handles convex polygons, as well as some
+ * non-convex polygons if we get lucky.
+ *
+ * Returns TRUE if the polygon was successfully rendered.  The rendering
+ * output is provided as callbacks (see the api).
+ */
+GLboolean __gl_renderCache( GLUtesselator *tess )
+{
+  CachedVertex *v0 = tess->cache;
+  CachedVertex *vn = v0 + tess->cacheCount;
+  CachedVertex *vc;
+  GLdouble norm[3];
+  int sign;
+
+  if( tess->cacheCount < 3 ) {
+    /* Degenerate contour -- no output */
+    return TRUE;
+  }
+
+  norm[0] = tess->normal[0];
+  norm[1] = tess->normal[1];
+  norm[2] = tess->normal[2];
+  if( norm[0] == 0 && norm[1] == 0 && norm[2] == 0 ) {
+    ComputeNormal( tess, norm, FALSE );
+  }
+
+  sign = ComputeNormal( tess, norm, TRUE );
+  if( sign == SIGN_INCONSISTENT ) {
+    /* Fan triangles did not have a consistent orientation */
+    return FALSE;
+  }
+  if( sign == 0 ) {
+    /* All triangles were degenerate */
+    return TRUE;
+  }
+
+  /* Make sure we do the right thing for each winding rule */
+  switch( tess->windingRule ) {
+  case GLU_TESS_WINDING_ODD:
+  case GLU_TESS_WINDING_NONZERO:
+    break;
+  case GLU_TESS_WINDING_POSITIVE:
+    if( sign < 0 ) return TRUE;
+    break;
+  case GLU_TESS_WINDING_NEGATIVE:
+    if( sign > 0 ) return TRUE;
+    break;
+  case GLU_TESS_WINDING_ABS_GEQ_TWO:
+    return TRUE;
+  }
+
+  CALL_BEGIN_OR_BEGIN_DATA( tess->boundaryOnly ? GL_LINE_LOOP
+			  : (tess->cacheCount > 3) ? GL_TRIANGLE_FAN
+			  : GL_TRIANGLES );
+
+  CALL_VERTEX_OR_VERTEX_DATA( v0->data ); 
+  if( sign > 0 ) {
+    for( vc = v0+1; vc < vn; ++vc ) {
+      CALL_VERTEX_OR_VERTEX_DATA( vc->data ); 
+    }
+  } else {
+    for( vc = vn-1; vc > v0; --vc ) {
+      CALL_VERTEX_OR_VERTEX_DATA( vc->data ); 
+    }
+  }
+  CALL_END_OR_END_DATA();
+  return TRUE;
+}
diff --git a/third_party/glu/libtess/render.h b/third_party/glu/libtess/render.h
new file mode 100644
index 0000000..598423e
--- /dev/null
+++ b/third_party/glu/libtess/render.h
@@ -0,0 +1,59 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/render.h#5 $
+*/
+
+#ifndef __render_h_
+#define __render_h_
+
+#include "mesh.h"
+
+/* __gl_renderMesh( tess, mesh ) takes a mesh and breaks it into triangle
+ * fans, strips, and separate triangles.  A substantial effort is made
+ * to use as few rendering primitives as possible (ie. to make the fans
+ * and strips as large as possible).
+ *
+ * The rendering output is provided as callbacks (see the api).
+ */
+void __gl_renderMesh( GLUtesselator *tess, GLUmesh *mesh );
+void __gl_renderBoundary( GLUtesselator *tess, GLUmesh *mesh );
+
+GLboolean __gl_renderCache( GLUtesselator *tess );
+
+#endif
diff --git a/third_party/glu/libtess/sweep.c b/third_party/glu/libtess/sweep.c
new file mode 100644
index 0000000..6fe8a47
--- /dev/null
+++ b/third_party/glu/libtess/sweep.c
@@ -0,0 +1,1362 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/sweep.c#5 $
+*/
+
+#include "gluos.h"
+#include <assert.h>
+#include <stddef.h>
+#include <setjmp.h>		/* longjmp */
+#include <limits.h>		/* LONG_MAX */
+
+#include "mesh.h"
+#include "geom.h"
+#include "tess.h"
+#include "dict.h"
+#include "priorityq.h"
+#include "memalloc.h"
+#include "sweep.h"
+
+#define TRUE 1
+#define FALSE 0
+
+#ifdef FOR_TRITE_TEST_PROGRAM
+extern void DebugEvent( GLUtesselator *tess );
+#else
+#define DebugEvent( tess )
+#endif
+
+/*
+ * Invariants for the Edge Dictionary.
+ * - each pair of adjacent edges e2=Succ(e1) satisfies EdgeLeq(e1,e2)
+ *   at any valid location of the sweep event
+ * - if EdgeLeq(e2,e1) as well (at any valid sweep event), then e1 and e2
+ *   share a common endpoint
+ * - for each e, e->Dst has been processed, but not e->Org
+ * - each edge e satisfies VertLeq(e->Dst,event) && VertLeq(event,e->Org)
+ *   where "event" is the current sweep line event.
+ * - no edge e has zero length
+ *
+ * Invariants for the Mesh (the processed portion).
+ * - the portion of the mesh left of the sweep line is a planar graph,
+ *   ie. there is *some* way to embed it in the plane
+ * - no processed edge has zero length
+ * - no two processed vertices have identical coordinates
+ * - each "inside" region is monotone, ie. can be broken into two chains
+ *   of monotonically increasing vertices according to VertLeq(v1,v2)
+ *   - a non-invariant: these chains may intersect (very slightly)
+ *
+ * Invariants for the Sweep.
+ * - if none of the edges incident to the event vertex have an activeRegion
+ *   (ie. none of these edges are in the edge dictionary), then the vertex
+ *   has only right-going edges.
+ * - if an edge is marked "fixUpperEdge" (it is a temporary edge introduced
+ *   by ConnectRightVertex), then it is the only right-going edge from
+ *   its associated vertex.  (This says that these edges exist only
+ *   when it is necessary.)
+ */
+
+#define MAX(x,y)	((x) >= (y) ? (x) : (y))
+#define MIN(x,y)	((x) <= (y) ? (x) : (y))
+
+/* When we merge two edges into one, we need to compute the combined
+ * winding of the new edge.
+ */
+#define AddWinding(eDst,eSrc)	(eDst->winding += eSrc->winding, \
+				 eDst->Sym->winding += eSrc->Sym->winding)
+
+static void SweepEvent( GLUtesselator *tess, GLUvertex *vEvent );
+static void WalkDirtyRegions( GLUtesselator *tess, ActiveRegion *regUp );
+static int CheckForRightSplice( GLUtesselator *tess, ActiveRegion *regUp );
+
+static int EdgeLeq( GLUtesselator *tess, ActiveRegion *reg1,
+		    ActiveRegion *reg2 )
+/*
+ * Both edges must be directed from right to left (this is the canonical
+ * direction for the upper edge of each region).
+ *
+ * The strategy is to evaluate a "t" value for each edge at the
+ * current sweep line position, given by tess->event.  The calculations
+ * are designed to be very stable, but of course they are not perfect.
+ *
+ * Special case: if both edge destinations are at the sweep event,
+ * we sort the edges by slope (they would otherwise compare equally).
+ */
+{
+  GLUvertex *event = tess->event;
+  GLUhalfEdge *e1, *e2;
+  GLdouble t1, t2;
+
+  e1 = reg1->eUp;
+  e2 = reg2->eUp;
+
+  if( e1->Dst == event ) {
+    if( e2->Dst == event ) {
+      /* Two edges right of the sweep line which meet at the sweep event.
+       * Sort them by slope.
+       */
+      if( VertLeq( e1->Org, e2->Org )) {
+	return EdgeSign( e2->Dst, e1->Org, e2->Org ) <= 0;
+      }
+      return EdgeSign( e1->Dst, e2->Org, e1->Org ) >= 0;
+    }
+    return EdgeSign( e2->Dst, event, e2->Org ) <= 0;
+  }
+  if( e2->Dst == event ) {
+    return EdgeSign( e1->Dst, event, e1->Org ) >= 0;
+  }
+
+  /* General case - compute signed distance *from* e1, e2 to event */
+  t1 = EdgeEval( e1->Dst, event, e1->Org );
+  t2 = EdgeEval( e2->Dst, event, e2->Org );
+  return (t1 >= t2);
+}
+
+
+static void DeleteRegion( GLUtesselator *tess, ActiveRegion *reg )
+{
+  if( reg->fixUpperEdge ) {
+    /* It was created with zero winding number, so it better be
+     * deleted with zero winding number (ie. it better not get merged
+     * with a real edge).
+     */
+    assert( reg->eUp->winding == 0 );
+  }
+  reg->eUp->activeRegion = NULL;
+  dictDelete( tess->dict, reg->nodeUp ); /* __gl_dictListDelete */
+  memFree( reg );
+}
+
+
+static int FixUpperEdge( ActiveRegion *reg, GLUhalfEdge *newEdge )
+/*
+ * Replace an upper edge which needs fixing (see ConnectRightVertex).
+ */
+{
+  assert( reg->fixUpperEdge );
+  if ( !__gl_meshDelete( reg->eUp ) ) return 0;
+  reg->fixUpperEdge = FALSE;
+  reg->eUp = newEdge;
+  newEdge->activeRegion = reg;
+
+  return 1; 
+}
+
+static ActiveRegion *TopLeftRegion( ActiveRegion *reg )
+{
+  GLUvertex *org = reg->eUp->Org;
+  GLUhalfEdge *e;
+
+  /* Find the region above the uppermost edge with the same origin */
+  do {
+    reg = RegionAbove( reg );
+  } while( reg->eUp->Org == org );
+
+  /* If the edge above was a temporary edge introduced by ConnectRightVertex,
+   * now is the time to fix it.
+   */
+  if( reg->fixUpperEdge ) {
+    e = __gl_meshConnect( RegionBelow(reg)->eUp->Sym, reg->eUp->Lnext );
+    if (e == NULL) return NULL;
+    if ( !FixUpperEdge( reg, e ) ) return NULL;
+    reg = RegionAbove( reg );
+  }
+  return reg;
+}
+
+static ActiveRegion *TopRightRegion( ActiveRegion *reg )
+{
+  GLUvertex *dst = reg->eUp->Dst;
+
+  /* Find the region above the uppermost edge with the same destination */
+  do {
+    reg = RegionAbove( reg );
+  } while( reg->eUp->Dst == dst );
+  return reg;
+}
+
+static ActiveRegion *AddRegionBelow( GLUtesselator *tess,
+				     ActiveRegion *regAbove,
+				     GLUhalfEdge *eNewUp )
+/*
+ * Add a new active region to the sweep line, *somewhere* below "regAbove"
+ * (according to where the new edge belongs in the sweep-line dictionary).
+ * The upper edge of the new region will be "eNewUp".
+ * Winding number and "inside" flag are not updated.
+ */
+{
+  ActiveRegion *regNew = (ActiveRegion *)memAlloc( sizeof( ActiveRegion ));
+  if (regNew == NULL) longjmp(tess->env,1);
+
+  regNew->eUp = eNewUp;
+  /* __gl_dictListInsertBefore */ 
+  regNew->nodeUp = dictInsertBefore( tess->dict, regAbove->nodeUp, regNew );
+  if (regNew->nodeUp == NULL) longjmp(tess->env,1);
+  regNew->fixUpperEdge = FALSE;
+  regNew->sentinel = FALSE;
+  regNew->dirty = FALSE;
+
+  eNewUp->activeRegion = regNew;
+  return regNew;
+}
+
+static GLboolean IsWindingInside( GLUtesselator *tess, int n )
+{
+  switch( tess->windingRule ) {
+  case GLU_TESS_WINDING_ODD:
+    return (n & 1);
+  case GLU_TESS_WINDING_NONZERO:
+    return (n != 0);
+  case GLU_TESS_WINDING_POSITIVE:
+    return (n > 0);
+  case GLU_TESS_WINDING_NEGATIVE:
+    return (n < 0);
+  case GLU_TESS_WINDING_ABS_GEQ_TWO:
+    return (n >= 2) || (n <= -2);
+  }
+  /*LINTED*/
+  assert( FALSE );
+  /*NOTREACHED*/
+  return 0;
+}
+
+
+static void ComputeWinding( GLUtesselator *tess, ActiveRegion *reg )
+{
+  reg->windingNumber = RegionAbove(reg)->windingNumber + reg->eUp->winding;
+  reg->inside = IsWindingInside( tess, reg->windingNumber );
+}
+
+
+static void FinishRegion( GLUtesselator *tess, ActiveRegion *reg )
+/*
+ * Delete a region from the sweep line.  This happens when the upper
+ * and lower chains of a region meet (at a vertex on the sweep line).
+ * The "inside" flag is copied to the appropriate mesh face (we could
+ * not do this before -- since the structure of the mesh is always
+ * changing, this face may not have even existed until now).
+ */
+{
+  GLUhalfEdge *e = reg->eUp;
+  GLUface *f = e->Lface;
+
+  f->inside = reg->inside;
+  f->anEdge = e;   /* optimization for __gl_meshTessellateMonoRegion() */
+  DeleteRegion( tess, reg );
+}
+
+
+static GLUhalfEdge *FinishLeftRegions( GLUtesselator *tess,
+	       ActiveRegion *regFirst, ActiveRegion *regLast )
+/*
+ * We are given a vertex with one or more left-going edges.  All affected
+ * edges should be in the edge dictionary.  Starting at regFirst->eUp,
+ * we walk down deleting all regions where both edges have the same
+ * origin vOrg.  At the same time we copy the "inside" flag from the
+ * active region to the face, since at this point each face will belong
+ * to at most one region (this was not necessarily true until this point
+ * in the sweep).  The walk stops at the region above regLast; if regLast
+ * is NULL we walk as far as possible.  At the same time we relink the
+ * mesh if necessary, so that the ordering of edges around vOrg is the
+ * same as in the dictionary.
+ */
+{
+  ActiveRegion *reg, *regPrev;
+  GLUhalfEdge *e, *ePrev;
+
+  regPrev = regFirst;
+  ePrev = regFirst->eUp;
+  while( regPrev != regLast ) {
+    regPrev->fixUpperEdge = FALSE;	/* placement was OK */
+    reg = RegionBelow( regPrev );
+    e = reg->eUp;
+    if( e->Org != ePrev->Org ) {
+      if( ! reg->fixUpperEdge ) {
+	/* Remove the last left-going edge.  Even though there are no further
+	 * edges in the dictionary with this origin, there may be further
+	 * such edges in the mesh (if we are adding left edges to a vertex
+	 * that has already been processed).  Thus it is important to call
+	 * FinishRegion rather than just DeleteRegion.
+	 */
+	FinishRegion( tess, regPrev );
+	break;
+      }
+      /* If the edge below was a temporary edge introduced by
+       * ConnectRightVertex, now is the time to fix it.
+       */
+      e = __gl_meshConnect( ePrev->Lprev, e->Sym );
+      if (e == NULL) longjmp(tess->env,1);
+      if ( !FixUpperEdge( reg, e ) ) longjmp(tess->env,1);
+    }
+
+    /* Relink edges so that ePrev->Onext == e */
+    if( ePrev->Onext != e ) {
+      if ( !__gl_meshSplice( e->Oprev, e ) ) longjmp(tess->env,1);
+      if ( !__gl_meshSplice( ePrev, e ) ) longjmp(tess->env,1);
+    }
+    FinishRegion( tess, regPrev );	/* may change reg->eUp */
+    ePrev = reg->eUp;
+    regPrev = reg;
+  }
+  return ePrev;
+}
+
+
+static void AddRightEdges( GLUtesselator *tess, ActiveRegion *regUp,
+       GLUhalfEdge *eFirst, GLUhalfEdge *eLast, GLUhalfEdge *eTopLeft,
+       GLboolean cleanUp )
+/*
+ * Purpose: insert right-going edges into the edge dictionary, and update
+ * winding numbers and mesh connectivity appropriately.  All right-going
+ * edges share a common origin vOrg.  Edges are inserted CCW starting at
+ * eFirst; the last edge inserted is eLast->Oprev.  If vOrg has any
+ * left-going edges already processed, then eTopLeft must be the edge
+ * such that an imaginary upward vertical segment from vOrg would be
+ * contained between eTopLeft->Oprev and eTopLeft; otherwise eTopLeft
+ * should be NULL.
+ */
+{
+  ActiveRegion *reg, *regPrev;
+  GLUhalfEdge *e, *ePrev;
+  int firstTime = TRUE;
+
+  /* Insert the new right-going edges in the dictionary */
+  e = eFirst;
+  do {
+    assert( VertLeq( e->Org, e->Dst ));
+    AddRegionBelow( tess, regUp, e->Sym );
+    e = e->Onext;
+  } while ( e != eLast );
+
+  /* Walk *all* right-going edges from e->Org, in the dictionary order,
+   * updating the winding numbers of each region, and re-linking the mesh
+   * edges to match the dictionary ordering (if necessary).
+   */
+  if( eTopLeft == NULL ) {
+    eTopLeft = RegionBelow( regUp )->eUp->Rprev;
+  }
+  regPrev = regUp;
+  ePrev = eTopLeft;
+  for( ;; ) {
+    reg = RegionBelow( regPrev );
+    e = reg->eUp->Sym;
+    if( e->Org != ePrev->Org ) break;
+
+    if( e->Onext != ePrev ) {
+      /* Unlink e from its current position, and relink below ePrev */
+      if ( !__gl_meshSplice( e->Oprev, e ) ) longjmp(tess->env,1);
+      if ( !__gl_meshSplice( ePrev->Oprev, e ) ) longjmp(tess->env,1);
+    }
+    /* Compute the winding number and "inside" flag for the new regions */
+    reg->windingNumber = regPrev->windingNumber - e->winding;
+    reg->inside = IsWindingInside( tess, reg->windingNumber );
+
+    /* Check for two outgoing edges with same slope -- process these
+     * before any intersection tests (see example in __gl_computeInterior).
+     */
+    regPrev->dirty = TRUE;
+    if( ! firstTime && CheckForRightSplice( tess, regPrev )) {
+      AddWinding( e, ePrev );
+      DeleteRegion( tess, regPrev );
+      if ( !__gl_meshDelete( ePrev ) ) longjmp(tess->env,1);
+    }
+    firstTime = FALSE;
+    regPrev = reg;
+    ePrev = e;
+  }
+  regPrev->dirty = TRUE;
+  assert( regPrev->windingNumber - e->winding == reg->windingNumber );
+
+  if( cleanUp ) {
+    /* Check for intersections between newly adjacent edges. */
+    WalkDirtyRegions( tess, regPrev );
+  }
+}
+
+
+static void CallCombine( GLUtesselator *tess, GLUvertex *isect,
+			 void *data[4], GLfloat weights[4], int needed )
+{
+  GLdouble coords[3];
+
+  /* Copy coord data in case the callback changes it. */
+  coords[0] = isect->coords[0];
+  coords[1] = isect->coords[1];
+  coords[2] = isect->coords[2];
+
+  isect->data = NULL;
+  CALL_COMBINE_OR_COMBINE_DATA( coords, data, weights, &isect->data );
+  if( isect->data == NULL ) {
+    if( ! needed ) {
+      isect->data = data[0];
+    } else if( ! tess->fatalError ) {
+      /* The only way fatal error is when two edges are found to intersect,
+       * but the user has not provided the callback necessary to handle
+       * generated intersection points.
+       */
+      CALL_ERROR_OR_ERROR_DATA( GLU_TESS_NEED_COMBINE_CALLBACK );
+      tess->fatalError = TRUE;
+    }
+  }
+}
+
+static void SpliceMergeVertices( GLUtesselator *tess, GLUhalfEdge *e1,
+				 GLUhalfEdge *e2 )
+/*
+ * Two vertices with idential coordinates are combined into one.
+ * e1->Org is kept, while e2->Org is discarded.
+ */
+{
+  void *data[4] = { NULL, NULL, NULL, NULL };
+  GLfloat weights[4] = { 0.5, 0.5, 0.0, 0.0 };
+
+  data[0] = e1->Org->data;
+  data[1] = e2->Org->data;
+  CallCombine( tess, e1->Org, data, weights, FALSE );
+  if ( !__gl_meshSplice( e1, e2 ) ) longjmp(tess->env,1); 
+}
+
+static void VertexWeights( GLUvertex *isect, GLUvertex *org, GLUvertex *dst,
+                           GLfloat *weights )
+/*
+ * Find some weights which describe how the intersection vertex is
+ * a linear combination of "org" and "dest".  Each of the two edges
+ * which generated "isect" is allocated 50% of the weight; each edge
+ * splits the weight between its org and dst according to the
+ * relative distance to "isect".
+ */
+{
+  GLdouble t1 = VertL1dist( org, isect );
+  GLdouble t2 = VertL1dist( dst, isect );
+
+  weights[0] = 0.5 * t2 / (t1 + t2);
+  weights[1] = 0.5 * t1 / (t1 + t2);
+  isect->coords[0] += weights[0]*org->coords[0] + weights[1]*dst->coords[0];
+  isect->coords[1] += weights[0]*org->coords[1] + weights[1]*dst->coords[1];
+  isect->coords[2] += weights[0]*org->coords[2] + weights[1]*dst->coords[2];
+}
+
+
+static void GetIntersectData( GLUtesselator *tess, GLUvertex *isect,
+       GLUvertex *orgUp, GLUvertex *dstUp,
+       GLUvertex *orgLo, GLUvertex *dstLo )
+/*
+ * We've computed a new intersection point, now we need a "data" pointer
+ * from the user so that we can refer to this new vertex in the
+ * rendering callbacks.
+ */
+{
+  void *data[4];
+  GLfloat weights[4];
+
+  data[0] = orgUp->data;
+  data[1] = dstUp->data;
+  data[2] = orgLo->data;
+  data[3] = dstLo->data;
+
+  isect->coords[0] = isect->coords[1] = isect->coords[2] = 0;
+  VertexWeights( isect, orgUp, dstUp, &weights[0] );
+  VertexWeights( isect, orgLo, dstLo, &weights[2] );
+
+  CallCombine( tess, isect, data, weights, TRUE );
+}
+
+static int CheckForRightSplice( GLUtesselator *tess, ActiveRegion *regUp )
+/*
+ * Check the upper and lower edge of "regUp", to make sure that the
+ * eUp->Org is above eLo, or eLo->Org is below eUp (depending on which
+ * origin is leftmost).
+ *
+ * The main purpose is to splice right-going edges with the same
+ * dest vertex and nearly identical slopes (ie. we can't distinguish
+ * the slopes numerically).  However the splicing can also help us
+ * to recover from numerical errors.  For example, suppose at one
+ * point we checked eUp and eLo, and decided that eUp->Org is barely
+ * above eLo.  Then later, we split eLo into two edges (eg. from
+ * a splice operation like this one).  This can change the result of
+ * our test so that now eUp->Org is incident to eLo, or barely below it.
+ * We must correct this condition to maintain the dictionary invariants.
+ *
+ * One possibility is to check these edges for intersection again
+ * (ie. CheckForIntersect).  This is what we do if possible.  However
+ * CheckForIntersect requires that tess->event lies between eUp and eLo,
+ * so that it has something to fall back on when the intersection
+ * calculation gives us an unusable answer.  So, for those cases where
+ * we can't check for intersection, this routine fixes the problem
+ * by just splicing the offending vertex into the other edge.
+ * This is a guaranteed solution, no matter how degenerate things get.
+ * Basically this is a combinatorial solution to a numerical problem.
+ */
+{
+  ActiveRegion *regLo = RegionBelow(regUp);
+  GLUhalfEdge *eUp = regUp->eUp;
+  GLUhalfEdge *eLo = regLo->eUp;
+
+  if( VertLeq( eUp->Org, eLo->Org )) {
+    if( EdgeSign( eLo->Dst, eUp->Org, eLo->Org ) > 0 ) return FALSE;
+
+    /* eUp->Org appears to be below eLo */
+    if( ! VertEq( eUp->Org, eLo->Org )) {
+      /* Splice eUp->Org into eLo */
+      if ( __gl_meshSplitEdge( eLo->Sym ) == NULL) longjmp(tess->env,1);
+      if ( !__gl_meshSplice( eUp, eLo->Oprev ) ) longjmp(tess->env,1);
+      regUp->dirty = regLo->dirty = TRUE;
+
+    } else if( eUp->Org != eLo->Org ) {
+      /* merge the two vertices, discarding eUp->Org */
+      pqDelete( tess->pq, eUp->Org->pqHandle ); /* __gl_pqSortDelete */
+      SpliceMergeVertices( tess, eLo->Oprev, eUp );
+    }
+  } else {
+    if( EdgeSign( eUp->Dst, eLo->Org, eUp->Org ) < 0 ) return FALSE;
+
+    /* eLo->Org appears to be above eUp, so splice eLo->Org into eUp */
+    RegionAbove(regUp)->dirty = regUp->dirty = TRUE;
+    if (__gl_meshSplitEdge( eUp->Sym ) == NULL) longjmp(tess->env,1);
+    if ( !__gl_meshSplice( eLo->Oprev, eUp ) ) longjmp(tess->env,1);
+  }
+  return TRUE;
+}
+
+static int CheckForLeftSplice( GLUtesselator *tess, ActiveRegion *regUp )
+/*
+ * Check the upper and lower edge of "regUp", to make sure that the
+ * eUp->Dst is above eLo, or eLo->Dst is below eUp (depending on which
+ * destination is rightmost).
+ *
+ * Theoretically, this should always be true.  However, splitting an edge
+ * into two pieces can change the results of previous tests.  For example,
+ * suppose at one point we checked eUp and eLo, and decided that eUp->Dst
+ * is barely above eLo.  Then later, we split eLo into two edges (eg. from
+ * a splice operation like this one).  This can change the result of
+ * the test so that now eUp->Dst is incident to eLo, or barely below it.
+ * We must correct this condition to maintain the dictionary invariants
+ * (otherwise new edges might get inserted in the wrong place in the
+ * dictionary, and bad stuff will happen).
+ *
+ * We fix the problem by just splicing the offending vertex into the
+ * other edge.
+ */
+{
+  ActiveRegion *regLo = RegionBelow(regUp);
+  GLUhalfEdge *eUp = regUp->eUp;
+  GLUhalfEdge *eLo = regLo->eUp;
+  GLUhalfEdge *e;
+
+  assert( ! VertEq( eUp->Dst, eLo->Dst ));
+
+  if( VertLeq( eUp->Dst, eLo->Dst )) {
+    if( EdgeSign( eUp->Dst, eLo->Dst, eUp->Org ) < 0 ) return FALSE;
+
+    /* eLo->Dst is above eUp, so splice eLo->Dst into eUp */
+    RegionAbove(regUp)->dirty = regUp->dirty = TRUE;
+    e = __gl_meshSplitEdge( eUp );
+    if (e == NULL) longjmp(tess->env,1);
+    if ( !__gl_meshSplice( eLo->Sym, e ) ) longjmp(tess->env,1);
+    e->Lface->inside = regUp->inside;
+  } else {
+    if( EdgeSign( eLo->Dst, eUp->Dst, eLo->Org ) > 0 ) return FALSE;
+
+    /* eUp->Dst is below eLo, so splice eUp->Dst into eLo */
+    regUp->dirty = regLo->dirty = TRUE;
+    e = __gl_meshSplitEdge( eLo );
+    if (e == NULL) longjmp(tess->env,1);    
+    if ( !__gl_meshSplice( eUp->Lnext, eLo->Sym ) ) longjmp(tess->env,1);
+    e->Rface->inside = regUp->inside;
+  }
+  return TRUE;
+}
+
+
+static int CheckForIntersect( GLUtesselator *tess, ActiveRegion *regUp )
+/*
+ * Check the upper and lower edges of the given region to see if
+ * they intersect.  If so, create the intersection and add it
+ * to the data structures.
+ *
+ * Returns TRUE if adding the new intersection resulted in a recursive
+ * call to AddRightEdges(); in this case all "dirty" regions have been
+ * checked for intersections, and possibly regUp has been deleted.
+ */
+{
+  ActiveRegion *regLo = RegionBelow(regUp);
+  GLUhalfEdge *eUp = regUp->eUp;
+  GLUhalfEdge *eLo = regLo->eUp;
+  GLUvertex *orgUp = eUp->Org;
+  GLUvertex *orgLo = eLo->Org;
+  GLUvertex *dstUp = eUp->Dst;
+  GLUvertex *dstLo = eLo->Dst;
+  GLdouble tMinUp, tMaxLo;
+  GLUvertex isect, *orgMin;
+  GLUhalfEdge *e;
+
+  assert( ! VertEq( dstLo, dstUp ));
+  assert( EdgeSign( dstUp, tess->event, orgUp ) <= 0 );
+  assert( EdgeSign( dstLo, tess->event, orgLo ) >= 0 );
+  assert( orgUp != tess->event && orgLo != tess->event );
+  assert( ! regUp->fixUpperEdge && ! regLo->fixUpperEdge );
+
+  if( orgUp == orgLo ) return FALSE;	/* right endpoints are the same */
+
+  tMinUp = MIN( orgUp->t, dstUp->t );
+  tMaxLo = MAX( orgLo->t, dstLo->t );
+  if( tMinUp > tMaxLo ) return FALSE;	/* t ranges do not overlap */
+
+  if( VertLeq( orgUp, orgLo )) {
+    if( EdgeSign( dstLo, orgUp, orgLo ) > 0 ) return FALSE;
+  } else {
+    if( EdgeSign( dstUp, orgLo, orgUp ) < 0 ) return FALSE;
+  }
+
+  /* At this point the edges intersect, at least marginally */
+  DebugEvent( tess );
+
+  __gl_edgeIntersect( dstUp, orgUp, dstLo, orgLo, &isect );
+  /* The following properties are guaranteed: */
+  assert( MIN( orgUp->t, dstUp->t ) <= isect.t );
+  assert( isect.t <= MAX( orgLo->t, dstLo->t ));
+  assert( MIN( dstLo->s, dstUp->s ) <= isect.s );
+  assert( isect.s <= MAX( orgLo->s, orgUp->s ));
+
+  if( VertLeq( &isect, tess->event )) {
+    /* The intersection point lies slightly to the left of the sweep line,
+     * so move it until it''s slightly to the right of the sweep line.
+     * (If we had perfect numerical precision, this would never happen
+     * in the first place).  The easiest and safest thing to do is
+     * replace the intersection by tess->event.
+     */
+    isect.s = tess->event->s;
+    isect.t = tess->event->t;
+  }
+  /* Similarly, if the computed intersection lies to the right of the
+   * rightmost origin (which should rarely happen), it can cause
+   * unbelievable inefficiency on sufficiently degenerate inputs.
+   * (If you have the test program, try running test54.d with the
+   * "X zoom" option turned on).
+   */
+  orgMin = VertLeq( orgUp, orgLo ) ? orgUp : orgLo;
+  if( VertLeq( orgMin, &isect )) {
+    isect.s = orgMin->s;
+    isect.t = orgMin->t;
+  }
+
+  if( VertEq( &isect, orgUp ) || VertEq( &isect, orgLo )) {
+    /* Easy case -- intersection at one of the right endpoints */
+    (void) CheckForRightSplice( tess, regUp );
+    return FALSE;
+  }
+
+  if(    (! VertEq( dstUp, tess->event )
+	  && EdgeSign( dstUp, tess->event, &isect ) >= 0)
+      || (! VertEq( dstLo, tess->event )
+	  && EdgeSign( dstLo, tess->event, &isect ) <= 0 ))
+  {
+    /* Very unusual -- the new upper or lower edge would pass on the
+     * wrong side of the sweep event, or through it.  This can happen
+     * due to very small numerical errors in the intersection calculation.
+     */
+    if( dstLo == tess->event ) {
+      /* Splice dstLo into eUp, and process the new region(s) */
+      if (__gl_meshSplitEdge( eUp->Sym ) == NULL) longjmp(tess->env,1);
+      if ( !__gl_meshSplice( eLo->Sym, eUp ) ) longjmp(tess->env,1);
+      regUp = TopLeftRegion( regUp );
+      if (regUp == NULL) longjmp(tess->env,1);
+      eUp = RegionBelow(regUp)->eUp;
+      FinishLeftRegions( tess, RegionBelow(regUp), regLo );
+      AddRightEdges( tess, regUp, eUp->Oprev, eUp, eUp, TRUE );
+      return TRUE;
+    }
+    if( dstUp == tess->event ) {
+      /* Splice dstUp into eLo, and process the new region(s) */
+      if (__gl_meshSplitEdge( eLo->Sym ) == NULL) longjmp(tess->env,1);
+      if ( !__gl_meshSplice( eUp->Lnext, eLo->Oprev ) ) longjmp(tess->env,1); 
+      regLo = regUp;
+      regUp = TopRightRegion( regUp );
+      e = RegionBelow(regUp)->eUp->Rprev;
+      regLo->eUp = eLo->Oprev;
+      eLo = FinishLeftRegions( tess, regLo, NULL );
+      AddRightEdges( tess, regUp, eLo->Onext, eUp->Rprev, e, TRUE );
+      return TRUE;
+    }
+    /* Special case: called from ConnectRightVertex.  If either
+     * edge passes on the wrong side of tess->event, split it
+     * (and wait for ConnectRightVertex to splice it appropriately).
+     */
+    if( EdgeSign( dstUp, tess->event, &isect ) >= 0 ) {
+      RegionAbove(regUp)->dirty = regUp->dirty = TRUE;
+      if (__gl_meshSplitEdge( eUp->Sym ) == NULL) longjmp(tess->env,1);
+      eUp->Org->s = tess->event->s;
+      eUp->Org->t = tess->event->t;
+    }
+    if( EdgeSign( dstLo, tess->event, &isect ) <= 0 ) {
+      regUp->dirty = regLo->dirty = TRUE;
+      if (__gl_meshSplitEdge( eLo->Sym ) == NULL) longjmp(tess->env,1);
+      eLo->Org->s = tess->event->s;
+      eLo->Org->t = tess->event->t;
+    }
+    /* leave the rest for ConnectRightVertex */
+    return FALSE;
+  }
+
+  /* General case -- split both edges, splice into new vertex.
+   * When we do the splice operation, the order of the arguments is
+   * arbitrary as far as correctness goes.  However, when the operation
+   * creates a new face, the work done is proportional to the size of
+   * the new face.  We expect the faces in the processed part of
+   * the mesh (ie. eUp->Lface) to be smaller than the faces in the
+   * unprocessed original contours (which will be eLo->Oprev->Lface).
+   */
+  if (__gl_meshSplitEdge( eUp->Sym ) == NULL) longjmp(tess->env,1);
+  if (__gl_meshSplitEdge( eLo->Sym ) == NULL) longjmp(tess->env,1);
+  if ( !__gl_meshSplice( eLo->Oprev, eUp ) ) longjmp(tess->env,1);
+  eUp->Org->s = isect.s;
+  eUp->Org->t = isect.t;
+  eUp->Org->pqHandle = pqInsert( tess->pq, eUp->Org ); /* __gl_pqSortInsert */
+  if (eUp->Org->pqHandle == LONG_MAX) {
+     pqDeletePriorityQ(tess->pq);	/* __gl_pqSortDeletePriorityQ */
+     tess->pq = NULL;
+     longjmp(tess->env,1);
+  }
+  GetIntersectData( tess, eUp->Org, orgUp, dstUp, orgLo, dstLo );
+  RegionAbove(regUp)->dirty = regUp->dirty = regLo->dirty = TRUE;
+  return FALSE;
+}
+
+static void WalkDirtyRegions( GLUtesselator *tess, ActiveRegion *regUp )
+/*
+ * When the upper or lower edge of any region changes, the region is
+ * marked "dirty".  This routine walks through all the dirty regions
+ * and makes sure that the dictionary invariants are satisfied
+ * (see the comments at the beginning of this file).  Of course
+ * new dirty regions can be created as we make changes to restore
+ * the invariants.
+ */
+{
+  ActiveRegion *regLo = RegionBelow(regUp);
+  GLUhalfEdge *eUp, *eLo;
+
+  for( ;; ) {
+    /* Find the lowest dirty region (we walk from the bottom up). */
+    while( regLo->dirty ) {
+      regUp = regLo;
+      regLo = RegionBelow(regLo);
+    }
+    if( ! regUp->dirty ) {
+      regLo = regUp;
+      regUp = RegionAbove( regUp );
+      if( regUp == NULL || ! regUp->dirty ) {
+	/* We've walked all the dirty regions */
+	return;
+      }
+    }
+    regUp->dirty = FALSE;
+    eUp = regUp->eUp;
+    eLo = regLo->eUp;
+
+    if( eUp->Dst != eLo->Dst ) {
+      /* Check that the edge ordering is obeyed at the Dst vertices. */
+      if( CheckForLeftSplice( tess, regUp )) {
+
+	/* If the upper or lower edge was marked fixUpperEdge, then
+	 * we no longer need it (since these edges are needed only for
+	 * vertices which otherwise have no right-going edges).
+	 */
+	if( regLo->fixUpperEdge ) {
+	  DeleteRegion( tess, regLo );
+	  if ( !__gl_meshDelete( eLo ) ) longjmp(tess->env,1);
+	  regLo = RegionBelow( regUp );
+	  eLo = regLo->eUp;
+	} else if( regUp->fixUpperEdge ) {
+	  DeleteRegion( tess, regUp );
+	  if ( !__gl_meshDelete( eUp ) ) longjmp(tess->env,1);
+	  regUp = RegionAbove( regLo );
+	  eUp = regUp->eUp;
+	}
+      }
+    }
+    if( eUp->Org != eLo->Org ) {
+      if(    eUp->Dst != eLo->Dst
+	  && ! regUp->fixUpperEdge && ! regLo->fixUpperEdge
+          && (eUp->Dst == tess->event || eLo->Dst == tess->event) )
+      {
+	/* When all else fails in CheckForIntersect(), it uses tess->event
+	 * as the intersection location.  To make this possible, it requires
+	 * that tess->event lie between the upper and lower edges, and also
+	 * that neither of these is marked fixUpperEdge (since in the worst
+	 * case it might splice one of these edges into tess->event, and
+	 * violate the invariant that fixable edges are the only right-going
+	 * edge from their associated vertex).
+         */
+	if( CheckForIntersect( tess, regUp )) {
+	  /* WalkDirtyRegions() was called recursively; we're done */
+	  return;
+	}
+      } else {
+	/* Even though we can't use CheckForIntersect(), the Org vertices
+	 * may violate the dictionary edge ordering.  Check and correct this.
+	 */
+	(void) CheckForRightSplice( tess, regUp );
+      }
+    }
+    if( eUp->Org == eLo->Org && eUp->Dst == eLo->Dst ) {
+      /* A degenerate loop consisting of only two edges -- delete it. */
+      AddWinding( eLo, eUp );
+      DeleteRegion( tess, regUp );
+      if ( !__gl_meshDelete( eUp ) ) longjmp(tess->env,1);
+      regUp = RegionAbove( regLo );
+    }
+  }
+}
+
+
+static void ConnectRightVertex( GLUtesselator *tess, ActiveRegion *regUp,
+			        GLUhalfEdge *eBottomLeft )
+/*
+ * Purpose: connect a "right" vertex vEvent (one where all edges go left)
+ * to the unprocessed portion of the mesh.  Since there are no right-going
+ * edges, two regions (one above vEvent and one below) are being merged
+ * into one.  "regUp" is the upper of these two regions.
+ *
+ * There are two reasons for doing this (adding a right-going edge):
+ *  - if the two regions being merged are "inside", we must add an edge
+ *    to keep them separated (the combined region would not be monotone).
+ *  - in any case, we must leave some record of vEvent in the dictionary,
+ *    so that we can merge vEvent with features that we have not seen yet.
+ *    For example, maybe there is a vertical edge which passes just to
+ *    the right of vEvent; we would like to splice vEvent into this edge.
+ *
+ * However, we don't want to connect vEvent to just any vertex.  We don''t
+ * want the new edge to cross any other edges; otherwise we will create
+ * intersection vertices even when the input data had no self-intersections.
+ * (This is a bad thing; if the user's input data has no intersections,
+ * we don't want to generate any false intersections ourselves.)
+ *
+ * Our eventual goal is to connect vEvent to the leftmost unprocessed
+ * vertex of the combined region (the union of regUp and regLo).
+ * But because of unseen vertices with all right-going edges, and also
+ * new vertices which may be created by edge intersections, we don''t
+ * know where that leftmost unprocessed vertex is.  In the meantime, we
+ * connect vEvent to the closest vertex of either chain, and mark the region
+ * as "fixUpperEdge".  This flag says to delete and reconnect this edge
+ * to the next processed vertex on the boundary of the combined region.
+ * Quite possibly the vertex we connected to will turn out to be the
+ * closest one, in which case we won''t need to make any changes.
+ */
+{
+  GLUhalfEdge *eNew;
+  GLUhalfEdge *eTopLeft = eBottomLeft->Onext;
+  ActiveRegion *regLo = RegionBelow(regUp);
+  GLUhalfEdge *eUp = regUp->eUp;
+  GLUhalfEdge *eLo = regLo->eUp;
+  int degenerate = FALSE;
+
+  if( eUp->Dst != eLo->Dst ) {
+    (void) CheckForIntersect( tess, regUp );
+  }
+
+  /* Possible new degeneracies: upper or lower edge of regUp may pass
+   * through vEvent, or may coincide with new intersection vertex
+   */
+  if( VertEq( eUp->Org, tess->event )) {
+    if ( !__gl_meshSplice( eTopLeft->Oprev, eUp ) ) longjmp(tess->env,1);
+    regUp = TopLeftRegion( regUp );
+    if (regUp == NULL) longjmp(tess->env,1);
+    eTopLeft = RegionBelow( regUp )->eUp;
+    FinishLeftRegions( tess, RegionBelow(regUp), regLo );
+    degenerate = TRUE;
+  }
+  if( VertEq( eLo->Org, tess->event )) {
+    if ( !__gl_meshSplice( eBottomLeft, eLo->Oprev ) ) longjmp(tess->env,1);
+    eBottomLeft = FinishLeftRegions( tess, regLo, NULL );
+    degenerate = TRUE;
+  }
+  if( degenerate ) {
+    AddRightEdges( tess, regUp, eBottomLeft->Onext, eTopLeft, eTopLeft, TRUE );
+    return;
+  }
+
+  /* Non-degenerate situation -- need to add a temporary, fixable edge.
+   * Connect to the closer of eLo->Org, eUp->Org.
+   */
+  if( VertLeq( eLo->Org, eUp->Org )) {
+    eNew = eLo->Oprev;
+  } else {
+    eNew = eUp;
+  }
+  eNew = __gl_meshConnect( eBottomLeft->Lprev, eNew );
+  if (eNew == NULL) longjmp(tess->env,1);
+
+  /* Prevent cleanup, otherwise eNew might disappear before we've even
+   * had a chance to mark it as a temporary edge.
+   */
+  AddRightEdges( tess, regUp, eNew, eNew->Onext, eNew->Onext, FALSE );
+  eNew->Sym->activeRegion->fixUpperEdge = TRUE;
+  WalkDirtyRegions( tess, regUp );
+}
+
+/* Because vertices at exactly the same location are merged together
+ * before we process the sweep event, some degenerate cases can't occur.
+ * However if someone eventually makes the modifications required to
+ * merge features which are close together, the cases below marked
+ * TOLERANCE_NONZERO will be useful.  They were debugged before the
+ * code to merge identical vertices in the main loop was added.
+ */
+#define TOLERANCE_NONZERO	FALSE
+
+static void ConnectLeftDegenerate( GLUtesselator *tess,
+				   ActiveRegion *regUp, GLUvertex *vEvent )
+/*
+ * The event vertex lies exacty on an already-processed edge or vertex.
+ * Adding the new vertex involves splicing it into the already-processed
+ * part of the mesh.
+ */
+{
+  GLUhalfEdge *e, *eTopLeft, *eTopRight, *eLast;
+  ActiveRegion *reg;
+
+  e = regUp->eUp;
+  if( VertEq( e->Org, vEvent )) {
+    /* e->Org is an unprocessed vertex - just combine them, and wait
+     * for e->Org to be pulled from the queue
+     */
+    assert( TOLERANCE_NONZERO );
+    SpliceMergeVertices( tess, e, vEvent->anEdge );
+    return;
+  }
+  
+  if( ! VertEq( e->Dst, vEvent )) {
+    /* General case -- splice vEvent into edge e which passes through it */
+    if (__gl_meshSplitEdge( e->Sym ) == NULL) longjmp(tess->env,1);
+    if( regUp->fixUpperEdge ) {
+      /* This edge was fixable -- delete unused portion of original edge */
+      if ( !__gl_meshDelete( e->Onext ) ) longjmp(tess->env,1);
+      regUp->fixUpperEdge = FALSE;
+    }
+    if ( !__gl_meshSplice( vEvent->anEdge, e ) ) longjmp(tess->env,1);
+    SweepEvent( tess, vEvent );	/* recurse */
+    return;
+  }
+
+  /* vEvent coincides with e->Dst, which has already been processed.
+   * Splice in the additional right-going edges.
+   */
+  assert( TOLERANCE_NONZERO );
+  regUp = TopRightRegion( regUp );
+  reg = RegionBelow( regUp );
+  eTopRight = reg->eUp->Sym;
+  eTopLeft = eLast = eTopRight->Onext;
+  if( reg->fixUpperEdge ) {
+    /* Here e->Dst has only a single fixable edge going right.
+     * We can delete it since now we have some real right-going edges.
+     */
+    assert( eTopLeft != eTopRight );   /* there are some left edges too */
+    DeleteRegion( tess, reg );
+    if ( !__gl_meshDelete( eTopRight ) ) longjmp(tess->env,1);
+    eTopRight = eTopLeft->Oprev;
+  }
+  if ( !__gl_meshSplice( vEvent->anEdge, eTopRight ) ) longjmp(tess->env,1);
+  if( ! EdgeGoesLeft( eTopLeft )) {
+    /* e->Dst had no left-going edges -- indicate this to AddRightEdges() */
+    eTopLeft = NULL;
+  }
+  AddRightEdges( tess, regUp, eTopRight->Onext, eLast, eTopLeft, TRUE );
+}
+
+
+static void ConnectLeftVertex( GLUtesselator *tess, GLUvertex *vEvent )
+/*
+ * Purpose: connect a "left" vertex (one where both edges go right)
+ * to the processed portion of the mesh.  Let R be the active region
+ * containing vEvent, and let U and L be the upper and lower edge
+ * chains of R.  There are two possibilities:
+ *
+ * - the normal case: split R into two regions, by connecting vEvent to
+ *   the rightmost vertex of U or L lying to the left of the sweep line
+ *
+ * - the degenerate case: if vEvent is close enough to U or L, we
+ *   merge vEvent into that edge chain.  The subcases are:
+ *	- merging with the rightmost vertex of U or L
+ *	- merging with the active edge of U or L
+ *	- merging with an already-processed portion of U or L
+ */
+{
+  ActiveRegion *regUp, *regLo, *reg;
+  GLUhalfEdge *eUp, *eLo, *eNew;
+  ActiveRegion tmp;
+
+  /* assert( vEvent->anEdge->Onext->Onext == vEvent->anEdge ); */
+
+  /* Get a pointer to the active region containing vEvent */
+  tmp.eUp = vEvent->anEdge->Sym;
+  /* __GL_DICTLISTKEY */ /* __gl_dictListSearch */
+  regUp = (ActiveRegion *)dictKey( dictSearch( tess->dict, &tmp ));
+  regLo = RegionBelow( regUp );
+  eUp = regUp->eUp;
+  eLo = regLo->eUp;
+
+  /* Try merging with U or L first */
+  if( EdgeSign( eUp->Dst, vEvent, eUp->Org ) == 0 ) {
+    ConnectLeftDegenerate( tess, regUp, vEvent );
+    return;
+  }
+
+  /* Connect vEvent to rightmost processed vertex of either chain.
+   * e->Dst is the vertex that we will connect to vEvent.
+   */
+  reg = VertLeq( eLo->Dst, eUp->Dst ) ? regUp : regLo;
+
+  if( regUp->inside || reg->fixUpperEdge) {
+    if( reg == regUp ) {
+      eNew = __gl_meshConnect( vEvent->anEdge->Sym, eUp->Lnext );
+      if (eNew == NULL) longjmp(tess->env,1);
+    } else {
+      GLUhalfEdge *tempHalfEdge= __gl_meshConnect( eLo->Dnext, vEvent->anEdge);
+      if (tempHalfEdge == NULL) longjmp(tess->env,1);
+
+      eNew = tempHalfEdge->Sym;
+    }
+    if( reg->fixUpperEdge ) {
+      if ( !FixUpperEdge( reg, eNew ) ) longjmp(tess->env,1);
+    } else {
+      ComputeWinding( tess, AddRegionBelow( tess, regUp, eNew ));
+    }
+    SweepEvent( tess, vEvent );
+  } else {
+    /* The new vertex is in a region which does not belong to the polygon.
+     * We don''t need to connect this vertex to the rest of the mesh.
+     */
+    AddRightEdges( tess, regUp, vEvent->anEdge, vEvent->anEdge, NULL, TRUE );
+  }
+}
+
+
+static void SweepEvent( GLUtesselator *tess, GLUvertex *vEvent )
+/*
+ * Does everything necessary when the sweep line crosses a vertex.
+ * Updates the mesh and the edge dictionary.
+ */
+{
+  ActiveRegion *regUp, *reg;
+  GLUhalfEdge *e, *eTopLeft, *eBottomLeft;
+
+  tess->event = vEvent;		/* for access in EdgeLeq() */
+  DebugEvent( tess );
+  
+  /* Check if this vertex is the right endpoint of an edge that is
+   * already in the dictionary.  In this case we don't need to waste
+   * time searching for the location to insert new edges.
+   */
+  e = vEvent->anEdge;
+  while( e->activeRegion == NULL ) {
+    e = e->Onext;
+    if( e == vEvent->anEdge ) {
+      /* All edges go right -- not incident to any processed edges */
+      ConnectLeftVertex( tess, vEvent );
+      return;
+    }
+  }
+
+  /* Processing consists of two phases: first we "finish" all the
+   * active regions where both the upper and lower edges terminate
+   * at vEvent (ie. vEvent is closing off these regions).
+   * We mark these faces "inside" or "outside" the polygon according
+   * to their winding number, and delete the edges from the dictionary.
+   * This takes care of all the left-going edges from vEvent.
+   */
+  regUp = TopLeftRegion( e->activeRegion );
+  if (regUp == NULL) longjmp(tess->env,1);
+  reg = RegionBelow( regUp );
+  eTopLeft = reg->eUp;
+  eBottomLeft = FinishLeftRegions( tess, reg, NULL );
+
+  /* Next we process all the right-going edges from vEvent.  This
+   * involves adding the edges to the dictionary, and creating the
+   * associated "active regions" which record information about the
+   * regions between adjacent dictionary edges.
+   */
+  if( eBottomLeft->Onext == eTopLeft ) {
+    /* No right-going edges -- add a temporary "fixable" edge */
+    ConnectRightVertex( tess, regUp, eBottomLeft );
+  } else {
+    AddRightEdges( tess, regUp, eBottomLeft->Onext, eTopLeft, eTopLeft, TRUE );
+  }
+}
+
+
+/* Make the sentinel coordinates big enough that they will never be
+ * merged with real input features.  (Even with the largest possible
+ * input contour and the maximum tolerance of 1.0, no merging will be
+ * done with coordinates larger than 3 * GLU_TESS_MAX_COORD).
+ */
+#define SENTINEL_COORD	(4 * GLU_TESS_MAX_COORD)
+
+static void AddSentinel( GLUtesselator *tess, GLdouble t )
+/*
+ * We add two sentinel edges above and below all other edges,
+ * to avoid special cases at the top and bottom.
+ */
+{
+  GLUhalfEdge *e;
+  ActiveRegion *reg = (ActiveRegion *)memAlloc( sizeof( ActiveRegion ));
+  if (reg == NULL) longjmp(tess->env,1);
+
+  e = __gl_meshMakeEdge( tess->mesh );
+  if (e == NULL) longjmp(tess->env,1);
+
+  e->Org->s = SENTINEL_COORD;
+  e->Org->t = t;
+  e->Dst->s = -SENTINEL_COORD;
+  e->Dst->t = t;
+  tess->event = e->Dst;		/* initialize it */
+
+  reg->eUp = e;
+  reg->windingNumber = 0;
+  reg->inside = FALSE;
+  reg->fixUpperEdge = FALSE;
+  reg->sentinel = TRUE;
+  reg->dirty = FALSE;
+  reg->nodeUp = dictInsert( tess->dict, reg ); /* __gl_dictListInsertBefore */
+  if (reg->nodeUp == NULL) longjmp(tess->env,1);
+}
+
+
+static void InitEdgeDict( GLUtesselator *tess )
+/*
+ * We maintain an ordering of edge intersections with the sweep line.
+ * This order is maintained in a dynamic dictionary.
+ */
+{
+  /* __gl_dictListNewDict */
+  tess->dict = dictNewDict( tess, (int (*)(void *, DictKey, DictKey)) EdgeLeq );
+  if (tess->dict == NULL) longjmp(tess->env,1);
+
+  AddSentinel( tess, -SENTINEL_COORD );
+  AddSentinel( tess, SENTINEL_COORD );
+}
+
+
+static void DoneEdgeDict( GLUtesselator *tess )
+{
+  ActiveRegion *reg;
+#ifndef NDEBUG
+  int fixedEdges = 0;
+#endif
+
+  /* __GL_DICTLISTKEY */ /* __GL_DICTLISTMIN */
+  while( (reg = (ActiveRegion *)dictKey( dictMin( tess->dict ))) != NULL ) {
+    /*
+     * At the end of all processing, the dictionary should contain
+     * only the two sentinel edges, plus at most one "fixable" edge
+     * created by ConnectRightVertex().
+     */
+    if( ! reg->sentinel ) {
+      assert( reg->fixUpperEdge );
+      assert( ++fixedEdges == 1 );
+    }
+    assert( reg->windingNumber == 0 );
+    DeleteRegion( tess, reg );
+/*    __gl_meshDelete( reg->eUp );*/
+  }
+  dictDeleteDict( tess->dict );	/* __gl_dictListDeleteDict */
+}
+
+
+static void RemoveDegenerateEdges( GLUtesselator *tess )
+/*
+ * Remove zero-length edges, and contours with fewer than 3 vertices.
+ */
+{
+  GLUhalfEdge *e, *eNext, *eLnext;
+  GLUhalfEdge *eHead = &tess->mesh->eHead;
+
+  /*LINTED*/
+  for( e = eHead->next; e != eHead; e = eNext ) {
+    eNext = e->next;
+    eLnext = e->Lnext;
+    
+    if( VertEq( e->Org, e->Dst ) && e->Lnext->Lnext != e ) {
+      /* Zero-length edge, contour has at least 3 edges */
+      
+      SpliceMergeVertices( tess, eLnext, e );	/* deletes e->Org */
+      if ( !__gl_meshDelete( e ) ) longjmp(tess->env,1); /* e is a self-loop */
+      e = eLnext;
+      eLnext = e->Lnext;
+    }
+    if( eLnext->Lnext == e ) {
+      /* Degenerate contour (one or two edges) */
+      
+      if( eLnext != e ) {
+	if( eLnext == eNext || eLnext == eNext->Sym ) { eNext = eNext->next; }
+	if ( !__gl_meshDelete( eLnext ) ) longjmp(tess->env,1);
+      }
+      if( e == eNext || e == eNext->Sym ) { eNext = eNext->next; }
+      if ( !__gl_meshDelete( e ) ) longjmp(tess->env,1);
+    }
+  }
+}
+
+static int InitPriorityQ( GLUtesselator *tess )
+/*
+ * Insert all vertices into the priority queue which determines the
+ * order in which vertices cross the sweep line.
+ */
+{
+  PriorityQ *pq;
+  GLUvertex *v, *vHead;
+
+  /* __gl_pqSortNewPriorityQ */
+  pq = tess->pq = pqNewPriorityQ( (int (*)(PQkey, PQkey)) __gl_vertLeq );
+  if (pq == NULL) return 0;
+
+  vHead = &tess->mesh->vHead;
+  for( v = vHead->next; v != vHead; v = v->next ) {
+    v->pqHandle = pqInsert( pq, v ); /* __gl_pqSortInsert */
+    if (v->pqHandle == LONG_MAX) break;
+  }
+  if (v != vHead || !pqInit( pq ) ) { /* __gl_pqSortInit */
+    pqDeletePriorityQ(tess->pq);	/* __gl_pqSortDeletePriorityQ */
+    tess->pq = NULL;
+    return 0;
+  }
+
+  return 1;
+}
+
+
+static void DonePriorityQ( GLUtesselator *tess )
+{
+  pqDeletePriorityQ( tess->pq ); /* __gl_pqSortDeletePriorityQ */
+}
+
+
+static int RemoveDegenerateFaces( GLUmesh *mesh )
+/*
+ * Delete any degenerate faces with only two edges.  WalkDirtyRegions()
+ * will catch almost all of these, but it won't catch degenerate faces
+ * produced by splice operations on already-processed edges.
+ * The two places this can happen are in FinishLeftRegions(), when
+ * we splice in a "temporary" edge produced by ConnectRightVertex(),
+ * and in CheckForLeftSplice(), where we splice already-processed
+ * edges to ensure that our dictionary invariants are not violated
+ * by numerical errors.
+ *
+ * In both these cases it is *very* dangerous to delete the offending
+ * edge at the time, since one of the routines further up the stack
+ * will sometimes be keeping a pointer to that edge.
+ */
+{
+  GLUface *f, *fNext;
+  GLUhalfEdge *e;
+
+  /*LINTED*/
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = fNext ) {
+    fNext = f->next;
+    e = f->anEdge;
+    assert( e->Lnext != e );
+
+    if( e->Lnext->Lnext == e ) {
+      /* A face with only two edges */
+      AddWinding( e->Onext, e );
+      if ( !__gl_meshDelete( e ) ) return 0;
+    }
+  }
+  return 1;
+}
+
+int __gl_computeInterior( GLUtesselator *tess )
+/*
+ * __gl_computeInterior( tess ) computes the planar arrangement specified
+ * by the given contours, and further subdivides this arrangement
+ * into regions.  Each region is marked "inside" if it belongs
+ * to the polygon, according to the rule given by tess->windingRule.
+ * Each interior region is guaranteed be monotone.
+ */
+{
+  GLUvertex *v, *vNext;
+
+  tess->fatalError = FALSE;
+
+  /* Each vertex defines an event for our sweep line.  Start by inserting
+   * all the vertices in a priority queue.  Events are processed in
+   * lexicographic order, ie.
+   *
+   *	e1 < e2  iff  e1.x < e2.x || (e1.x == e2.x && e1.y < e2.y)
+   */
+  RemoveDegenerateEdges( tess );
+  if ( !InitPriorityQ( tess ) ) return 0; /* if error */
+  InitEdgeDict( tess );
+
+  /* __gl_pqSortExtractMin */
+  while( (v = (GLUvertex *)pqExtractMin( tess->pq )) != NULL ) {
+    for( ;; ) {
+      vNext = (GLUvertex *)pqMinimum( tess->pq ); /* __gl_pqSortMinimum */
+      if( vNext == NULL || ! VertEq( vNext, v )) break;
+      
+      /* Merge together all vertices at exactly the same location.
+       * This is more efficient than processing them one at a time,
+       * simplifies the code (see ConnectLeftDegenerate), and is also
+       * important for correct handling of certain degenerate cases.
+       * For example, suppose there are two identical edges A and B
+       * that belong to different contours (so without this code they would
+       * be processed by separate sweep events).  Suppose another edge C
+       * crosses A and B from above.  When A is processed, we split it
+       * at its intersection point with C.  However this also splits C,
+       * so when we insert B we may compute a slightly different
+       * intersection point.  This might leave two edges with a small
+       * gap between them.  This kind of error is especially obvious
+       * when using boundary extraction (GLU_TESS_BOUNDARY_ONLY).
+       */
+      vNext = (GLUvertex *)pqExtractMin( tess->pq ); /* __gl_pqSortExtractMin*/
+      SpliceMergeVertices( tess, v->anEdge, vNext->anEdge );
+    }
+    SweepEvent( tess, v );
+  }
+
+  /* Set tess->event for debugging purposes */
+  /* __GL_DICTLISTKEY */ /* __GL_DICTLISTMIN */ 
+  tess->event = ((ActiveRegion *) dictKey( dictMin( tess->dict )))->eUp->Org;
+  DebugEvent( tess );
+  DoneEdgeDict( tess );
+  DonePriorityQ( tess );
+
+  if ( !RemoveDegenerateFaces( tess->mesh ) ) return 0;
+  __gl_meshCheckMesh( tess->mesh );
+
+  return 1;
+}
diff --git a/third_party/glu/libtess/sweep.h b/third_party/glu/libtess/sweep.h
new file mode 100644
index 0000000..75d3e64
--- /dev/null
+++ b/third_party/glu/libtess/sweep.h
@@ -0,0 +1,84 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/sweep.h#5 $
+*/
+
+#ifndef __sweep_h_
+#define __sweep_h_
+
+#include "mesh.h"
+
+/* __gl_computeInterior( tess ) computes the planar arrangement specified
+ * by the given contours, and further subdivides this arrangement
+ * into regions.  Each region is marked "inside" if it belongs
+ * to the polygon, according to the rule given by tess->windingRule.
+ * Each interior region is guaranteed be monotone.
+ */
+int __gl_computeInterior( GLUtesselator *tess );
+
+
+/* The following is here *only* for access by debugging routines */
+
+#include "dict.h"
+
+/* For each pair of adjacent edges crossing the sweep line, there is
+ * an ActiveRegion to represent the region between them.  The active
+ * regions are kept in sorted order in a dynamic dictionary.  As the
+ * sweep line crosses each vertex, we update the affected regions.
+ */
+
+struct ActiveRegion {
+  GLUhalfEdge	*eUp;		/* upper edge, directed right to left */
+  DictNode	*nodeUp;	/* dictionary node corresponding to eUp */
+  int		windingNumber;	/* used to determine which regions are
+                                 * inside the polygon */
+  GLboolean	inside;		/* is this region inside the polygon? */
+  GLboolean	sentinel;	/* marks fake edges at t = +/-infinity */
+  GLboolean	dirty;		/* marks regions where the upper or lower
+                                 * edge has changed, but we haven't checked
+                                 * whether they intersect yet */
+  GLboolean	fixUpperEdge;	/* marks temporary edges introduced when
+                                 * we process a "right vertex" (one without
+                                 * any edges leaving to the right) */
+};
+
+#define RegionBelow(r)	((ActiveRegion *) dictKey(dictPred((r)->nodeUp)))
+#define RegionAbove(r)	((ActiveRegion *) dictKey(dictSucc((r)->nodeUp)))
+
+#endif
diff --git a/third_party/glu/libtess/tess.c b/third_party/glu/libtess/tess.c
new file mode 100644
index 0000000..c59a178
--- /dev/null
+++ b/third_party/glu/libtess/tess.c
@@ -0,0 +1,638 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/tess.c#7 $
+*/
+
+#include "gluos.h"
+#include <stddef.h>
+#include <assert.h>
+#include <setjmp.h>
+#include "memalloc.h"
+#include "tess.h"
+#include "mesh.h"
+#include "normal.h"
+#include "sweep.h"
+#include "tessmono.h"
+#include "render.h"
+
+#define GLU_TESS_DEFAULT_TOLERANCE 0.0
+#define GLU_TESS_MESH		100112	/* void (*)(GLUmesh *mesh)	    */
+
+#define TRUE 1
+#define FALSE 0
+
+/*ARGSUSED*/ static void GLAPIENTRY noBegin( GLenum type ) {}
+/*ARGSUSED*/ static void GLAPIENTRY noEdgeFlag( GLboolean boundaryEdge ) {}
+/*ARGSUSED*/ static void GLAPIENTRY noVertex( void *data ) {}
+/*ARGSUSED*/ static void GLAPIENTRY noEnd( void ) {}
+/*ARGSUSED*/ static void GLAPIENTRY noError( GLenum errnum ) {}
+/*ARGSUSED*/ static void GLAPIENTRY noCombine( GLdouble coords[3], void *data[4],
+                                    GLfloat weight[4], void **dataOut ) {}
+/*ARGSUSED*/ static void GLAPIENTRY noMesh( GLUmesh *mesh ) {}
+
+
+/*ARGSUSED*/ void GLAPIENTRY __gl_noBeginData( GLenum type,
+					     void *polygonData ) {}
+/*ARGSUSED*/ void GLAPIENTRY __gl_noEdgeFlagData( GLboolean boundaryEdge, 
+				       void *polygonData ) {}
+/*ARGSUSED*/ void GLAPIENTRY __gl_noVertexData( void *data,
+					      void *polygonData ) {}
+/*ARGSUSED*/ void GLAPIENTRY __gl_noEndData( void *polygonData ) {}
+/*ARGSUSED*/ void GLAPIENTRY __gl_noErrorData( GLenum errnum,
+					     void *polygonData ) {}
+/*ARGSUSED*/ void GLAPIENTRY __gl_noCombineData( GLdouble coords[3],
+					       void *data[4],
+					       GLfloat weight[4],
+					       void **outData,
+					       void *polygonData ) {}
+
+/* Half-edges are allocated in pairs (see mesh.c) */
+typedef struct { GLUhalfEdge e, eSym; } EdgePair;
+
+#define MAX(a,b)	((a) > (b) ? (a) : (b))
+#define MAX_FAST_ALLOC	(MAX(sizeof(EdgePair), \
+			 MAX(sizeof(GLUvertex),sizeof(GLUface))))
+
+
+GLUtesselator * GLAPIENTRY
+gluNewTess( void )
+{
+  GLUtesselator *tess;
+
+  /* Only initialize fields which can be changed by the api.  Other fields
+   * are initialized where they are used.
+   */
+
+  if (memInit( MAX_FAST_ALLOC ) == 0) {
+     return 0;			/* out of memory */
+  }
+  tess = (GLUtesselator *)memAlloc( sizeof( GLUtesselator ));
+  if (tess == NULL) {
+     return 0;			/* out of memory */
+  }
+
+  tess->state = T_DORMANT;
+
+  tess->normal[0] = 0;
+  tess->normal[1] = 0;
+  tess->normal[2] = 0;
+
+  tess->relTolerance = GLU_TESS_DEFAULT_TOLERANCE;
+  tess->windingRule = GLU_TESS_WINDING_ODD;
+  tess->flagBoundary = FALSE;
+  tess->boundaryOnly = FALSE;
+
+  tess->callBegin = &noBegin;
+  tess->callEdgeFlag = &noEdgeFlag;
+  tess->callVertex = &noVertex;
+  tess->callEnd = &noEnd;
+
+  tess->callError = &noError;
+  tess->callCombine = &noCombine;
+  tess->callMesh = &noMesh;
+
+  tess->callBeginData= &__gl_noBeginData;
+  tess->callEdgeFlagData= &__gl_noEdgeFlagData;
+  tess->callVertexData= &__gl_noVertexData;
+  tess->callEndData= &__gl_noEndData;
+  tess->callErrorData= &__gl_noErrorData;
+  tess->callCombineData= &__gl_noCombineData;
+
+  tess->polygonData= NULL;
+
+  return tess;
+}
+
+static void MakeDormant( GLUtesselator *tess )
+{
+  /* Return the tessellator to its original dormant state. */
+
+  if( tess->mesh != NULL ) {
+    __gl_meshDeleteMesh( tess->mesh );
+  }
+  tess->state = T_DORMANT;
+  tess->lastEdge = NULL;
+  tess->mesh = NULL;
+}
+
+#define RequireState( tess, s )   if( tess->state != s ) GotoState(tess,s)
+
+static void GotoState( GLUtesselator *tess, enum TessState newState )
+{
+  while( tess->state != newState ) {
+    /* We change the current state one level at a time, to get to
+     * the desired state.
+     */
+    if( tess->state < newState ) {
+      switch( tess->state ) {
+      case T_DORMANT:
+	CALL_ERROR_OR_ERROR_DATA( GLU_TESS_MISSING_BEGIN_POLYGON );
+	gluTessBeginPolygon( tess, NULL );
+	break;
+      case T_IN_POLYGON:
+	CALL_ERROR_OR_ERROR_DATA( GLU_TESS_MISSING_BEGIN_CONTOUR );
+	gluTessBeginContour( tess );
+	break;
+      default:
+        assert(0);
+        break;
+      }
+    } else {
+      switch( tess->state ) {
+      case T_IN_CONTOUR:
+	CALL_ERROR_OR_ERROR_DATA( GLU_TESS_MISSING_END_CONTOUR );
+	gluTessEndContour( tess );
+	break;
+      case T_IN_POLYGON:
+	CALL_ERROR_OR_ERROR_DATA( GLU_TESS_MISSING_END_POLYGON );
+	/* gluTessEndPolygon( tess ) is too much work! */
+	MakeDormant( tess );
+	break;
+      default:
+        assert(0);
+        break;
+      }
+    }
+  }
+}
+
+
+void GLAPIENTRY
+gluDeleteTess( GLUtesselator *tess )
+{
+  RequireState( tess, T_DORMANT );
+  memFree( tess );
+}
+
+
+void GLAPIENTRY
+gluTessProperty( GLUtesselator *tess, GLenum which, GLdouble value )
+{
+  GLenum windingRule;
+
+  switch( which ) {
+  case GLU_TESS_TOLERANCE:
+    if( value < 0.0 || value > 1.0 ) break;
+    tess->relTolerance = value;
+    return;
+
+  case GLU_TESS_WINDING_RULE:
+    windingRule = (GLenum) value;
+    if( windingRule != value ) break;	/* not an integer */
+
+    switch( windingRule ) {
+    case GLU_TESS_WINDING_ODD:
+    case GLU_TESS_WINDING_NONZERO:
+    case GLU_TESS_WINDING_POSITIVE:
+    case GLU_TESS_WINDING_NEGATIVE:
+    case GLU_TESS_WINDING_ABS_GEQ_TWO:
+      tess->windingRule = windingRule;
+      return;
+    default:
+      break;
+    }
+
+  case GLU_TESS_BOUNDARY_ONLY:
+    tess->boundaryOnly = (value != 0);
+    return;
+
+  default:
+    CALL_ERROR_OR_ERROR_DATA( GLU_INVALID_ENUM );
+    return;
+  }
+  CALL_ERROR_OR_ERROR_DATA( GLU_INVALID_VALUE );
+}
+
+/* Returns tessellator property */
+void GLAPIENTRY
+gluGetTessProperty( GLUtesselator *tess, GLenum which, GLdouble *value )
+{
+   switch (which) {
+   case GLU_TESS_TOLERANCE:
+      /* tolerance should be in range [0..1] */
+      assert(0.0 <= tess->relTolerance && tess->relTolerance <= 1.0);
+      *value= tess->relTolerance;
+      break;    
+   case GLU_TESS_WINDING_RULE:
+      assert(tess->windingRule == GLU_TESS_WINDING_ODD ||
+	     tess->windingRule == GLU_TESS_WINDING_NONZERO ||
+	     tess->windingRule == GLU_TESS_WINDING_POSITIVE ||
+	     tess->windingRule == GLU_TESS_WINDING_NEGATIVE ||
+	     tess->windingRule == GLU_TESS_WINDING_ABS_GEQ_TWO);
+      *value= tess->windingRule;
+      break;
+   case GLU_TESS_BOUNDARY_ONLY:
+      assert(tess->boundaryOnly == TRUE || tess->boundaryOnly == FALSE);
+      *value= tess->boundaryOnly;
+      break;
+   default:
+      *value= 0.0;
+      CALL_ERROR_OR_ERROR_DATA( GLU_INVALID_ENUM );
+      break;
+   }
+} /* gluGetTessProperty() */
+
+void GLAPIENTRY
+gluTessNormal( GLUtesselator *tess, GLdouble x, GLdouble y, GLdouble z )
+{
+  tess->normal[0] = x;
+  tess->normal[1] = y;
+  tess->normal[2] = z;
+}
+
+void GLAPIENTRY
+gluTessCallback( GLUtesselator *tess, GLenum which, void (GLAPIENTRY *fn)())
+{
+  switch( which ) {
+  case GLU_TESS_BEGIN:
+    tess->callBegin = (fn == NULL) ? &noBegin : (void (GLAPIENTRY *)(GLenum)) fn;
+    return;
+  case GLU_TESS_BEGIN_DATA:
+    tess->callBeginData = (fn == NULL) ?
+	&__gl_noBeginData : (void (GLAPIENTRY *)(GLenum, void *)) fn;
+    return;
+  case GLU_TESS_EDGE_FLAG:
+    tess->callEdgeFlag = (fn == NULL) ? &noEdgeFlag :
+					(void (GLAPIENTRY *)(GLboolean)) fn;
+    /* If the client wants boundary edges to be flagged,
+     * we render everything as separate triangles (no strips or fans).
+     */
+    tess->flagBoundary = (fn != NULL);
+    return;
+  case GLU_TESS_EDGE_FLAG_DATA:
+    tess->callEdgeFlagData= (fn == NULL) ?
+	&__gl_noEdgeFlagData : (void (GLAPIENTRY *)(GLboolean, void *)) fn; 
+    /* If the client wants boundary edges to be flagged,
+     * we render everything as separate triangles (no strips or fans).
+     */
+    tess->flagBoundary = (fn != NULL);
+    return;
+  case GLU_TESS_VERTEX:
+    tess->callVertex = (fn == NULL) ? &noVertex :
+				      (void (GLAPIENTRY *)(void *)) fn;
+    return;
+  case GLU_TESS_VERTEX_DATA:
+    tess->callVertexData = (fn == NULL) ?
+	&__gl_noVertexData : (void (GLAPIENTRY *)(void *, void *)) fn;
+    return;
+  case GLU_TESS_END:
+    tess->callEnd = (fn == NULL) ? &noEnd : (void (GLAPIENTRY *)(void)) fn;
+    return;
+  case GLU_TESS_END_DATA:
+    tess->callEndData = (fn == NULL) ? &__gl_noEndData : 
+                                       (void (GLAPIENTRY *)(void *)) fn;
+    return;
+  case GLU_TESS_ERROR:
+    tess->callError = (fn == NULL) ? &noError : (void (GLAPIENTRY *)(GLenum)) fn;
+    return;
+  case GLU_TESS_ERROR_DATA:
+    tess->callErrorData = (fn == NULL) ?
+	&__gl_noErrorData : (void (GLAPIENTRY *)(GLenum, void *)) fn;
+    return;
+  case GLU_TESS_COMBINE:
+    tess->callCombine = (fn == NULL) ? &noCombine :
+	(void (GLAPIENTRY *)(GLdouble [3],void *[4], GLfloat [4], void ** )) fn;
+    return;
+  case GLU_TESS_COMBINE_DATA:
+    tess->callCombineData = (fn == NULL) ? &__gl_noCombineData :
+                                           (void (GLAPIENTRY *)(GLdouble [3],
+						     void *[4], 
+						     GLfloat [4], 
+						     void **,
+						     void *)) fn;
+    return;
+  case GLU_TESS_MESH:
+    tess->callMesh = (fn == NULL) ? &noMesh : (void (GLAPIENTRY *)(GLUmesh *)) fn;
+    return;
+  default:
+    CALL_ERROR_OR_ERROR_DATA( GLU_INVALID_ENUM );
+    return;
+  }
+}
+
+static int AddVertex( GLUtesselator *tess, GLdouble coords[3], void *data )
+{
+  GLUhalfEdge *e;
+
+  e = tess->lastEdge;
+  if( e == NULL ) {
+    /* Make a self-loop (one vertex, one edge). */
+
+    e = __gl_meshMakeEdge( tess->mesh );
+    if (e == NULL) return 0;
+    if ( !__gl_meshSplice( e, e->Sym ) ) return 0;
+  } else {
+    /* Create a new vertex and edge which immediately follow e
+     * in the ordering around the left face.
+     */
+    if (__gl_meshSplitEdge( e ) == NULL) return 0;
+    e = e->Lnext;
+  }
+
+  /* The new vertex is now e->Org. */
+  e->Org->data = data;
+  e->Org->coords[0] = coords[0];
+  e->Org->coords[1] = coords[1];
+  e->Org->coords[2] = coords[2];
+  
+  /* The winding of an edge says how the winding number changes as we
+   * cross from the edge''s right face to its left face.  We add the
+   * vertices in such an order that a CCW contour will add +1 to
+   * the winding number of the region inside the contour.
+   */
+  e->winding = 1;
+  e->Sym->winding = -1;
+
+  tess->lastEdge = e;
+
+  return 1;
+}
+
+
+static void CacheVertex( GLUtesselator *tess, GLdouble coords[3], void *data )
+{
+  CachedVertex *v = &tess->cache[tess->cacheCount];
+
+  v->data = data;
+  v->coords[0] = coords[0];
+  v->coords[1] = coords[1];
+  v->coords[2] = coords[2];
+  ++tess->cacheCount;
+}
+
+
+static int EmptyCache( GLUtesselator *tess )
+{
+  CachedVertex *v = tess->cache;
+  CachedVertex *vLast;
+
+  tess->mesh = __gl_meshNewMesh();
+  if (tess->mesh == NULL) return 0;
+
+  for( vLast = v + tess->cacheCount; v < vLast; ++v ) {
+    if ( !AddVertex( tess, v->coords, v->data ) ) return 0;
+  }
+  tess->cacheCount = 0;
+  tess->emptyCache = FALSE;
+
+  return 1;
+}
+
+
+void GLAPIENTRY
+gluTessVertex( GLUtesselator *tess, GLdouble coords[3], void *data )
+{
+  int i, tooLarge = FALSE;
+  GLdouble x, clamped[3];
+
+  RequireState( tess, T_IN_CONTOUR );
+
+  if( tess->emptyCache ) {
+    if ( !EmptyCache( tess ) ) {
+       CALL_ERROR_OR_ERROR_DATA( GLU_OUT_OF_MEMORY );
+       return;
+    }
+    tess->lastEdge = NULL;
+  }
+  for( i = 0; i < 3; ++i ) {
+    x = coords[i];
+    if( x < - GLU_TESS_MAX_COORD ) {
+      x = - GLU_TESS_MAX_COORD;
+      tooLarge = TRUE;
+    }
+    if( x > GLU_TESS_MAX_COORD ) {
+      x = GLU_TESS_MAX_COORD;
+      tooLarge = TRUE;
+    }
+    clamped[i] = x;
+  }
+  if( tooLarge ) {
+    CALL_ERROR_OR_ERROR_DATA( GLU_TESS_COORD_TOO_LARGE );
+  }
+
+  if( tess->mesh == NULL ) {
+    if( tess->cacheCount < TESS_MAX_CACHE ) {
+      CacheVertex( tess, clamped, data );
+      return;
+    }
+    if ( !EmptyCache( tess ) ) {
+       CALL_ERROR_OR_ERROR_DATA( GLU_OUT_OF_MEMORY );
+       return;
+    }
+  }
+  if ( !AddVertex( tess, clamped, data ) ) {
+       CALL_ERROR_OR_ERROR_DATA( GLU_OUT_OF_MEMORY );
+  }
+}
+
+
+void GLAPIENTRY
+gluTessBeginPolygon( GLUtesselator *tess, void *data )
+{
+  RequireState( tess, T_DORMANT );
+
+  tess->state = T_IN_POLYGON;
+  tess->cacheCount = 0;
+  tess->emptyCache = FALSE;
+  tess->mesh = NULL;
+
+  tess->polygonData= data;
+}
+
+
+void GLAPIENTRY
+gluTessBeginContour( GLUtesselator *tess )
+{
+  RequireState( tess, T_IN_POLYGON );
+
+  tess->state = T_IN_CONTOUR;
+  tess->lastEdge = NULL;
+  if( tess->cacheCount > 0 ) {
+    /* Just set a flag so we don't get confused by empty contours
+     * -- these can be generated accidentally with the obsolete
+     * NextContour() interface.
+     */
+    tess->emptyCache = TRUE;
+  }
+}
+
+
+void GLAPIENTRY
+gluTessEndContour( GLUtesselator *tess )
+{
+  RequireState( tess, T_IN_CONTOUR );
+  tess->state = T_IN_POLYGON;
+}
+
+void GLAPIENTRY
+gluTessEndPolygon( GLUtesselator *tess )
+{
+  GLUmesh *mesh;
+
+  if (setjmp(tess->env) != 0) {	
+     /* come back here if out of memory */
+     CALL_ERROR_OR_ERROR_DATA( GLU_OUT_OF_MEMORY );
+     return;
+  }
+
+  RequireState( tess, T_IN_POLYGON );
+  tess->state = T_DORMANT;
+
+  if( tess->mesh == NULL ) {
+    if( ! tess->flagBoundary && tess->callMesh == &noMesh ) {
+
+      /* Try some special code to make the easy cases go quickly
+       * (eg. convex polygons).  This code does NOT handle multiple contours,
+       * intersections, edge flags, and of course it does not generate
+       * an explicit mesh either.
+       */
+      if( __gl_renderCache( tess )) {
+	tess->polygonData= NULL; 
+	return;
+      }
+    }
+    if ( !EmptyCache( tess ) ) longjmp(tess->env,1); /* could've used a label*/
+  }
+
+  /* Determine the polygon normal and project vertices onto the plane
+   * of the polygon.
+   */
+  __gl_projectPolygon( tess );
+
+  /* __gl_computeInterior( tess ) computes the planar arrangement specified
+   * by the given contours, and further subdivides this arrangement
+   * into regions.  Each region is marked "inside" if it belongs
+   * to the polygon, according to the rule given by tess->windingRule.
+   * Each interior region is guaranteed be monotone.
+   */
+  if ( !__gl_computeInterior( tess ) ) {
+     longjmp(tess->env,1);	/* could've used a label */
+  }
+
+  mesh = tess->mesh;
+  if( ! tess->fatalError ) {
+    int rc = 1;
+
+    /* If the user wants only the boundary contours, we throw away all edges
+     * except those which separate the interior from the exterior.
+     * Otherwise we tessellate all the regions marked "inside".
+     */
+    if( tess->boundaryOnly ) {
+      rc = __gl_meshSetWindingNumber( mesh, 1, TRUE );
+    } else {
+      rc = __gl_meshTessellateInterior( mesh ); 
+    }
+    if (rc == 0) longjmp(tess->env,1);	/* could've used a label */
+
+    __gl_meshCheckMesh( mesh );
+
+    if( tess->callBegin != &noBegin || tess->callEnd != &noEnd
+       || tess->callVertex != &noVertex || tess->callEdgeFlag != &noEdgeFlag 
+       || tess->callBeginData != &__gl_noBeginData 
+       || tess->callEndData != &__gl_noEndData
+       || tess->callVertexData != &__gl_noVertexData
+       || tess->callEdgeFlagData != &__gl_noEdgeFlagData )
+    {
+      if( tess->boundaryOnly ) {
+	__gl_renderBoundary( tess, mesh );  /* output boundary contours */
+      } else {
+	__gl_renderMesh( tess, mesh );	   /* output strips and fans */
+      }
+    }
+    if( tess->callMesh != &noMesh ) {
+
+      /* Throw away the exterior faces, so that all faces are interior.
+       * This way the user doesn't have to check the "inside" flag,
+       * and we don't need to even reveal its existence.  It also leaves
+       * the freedom for an implementation to not generate the exterior
+       * faces in the first place.
+       */
+      __gl_meshDiscardExterior( mesh );
+      (*tess->callMesh)( mesh );		/* user wants the mesh itself */
+      tess->mesh = NULL;
+      tess->polygonData= NULL;
+      return;
+    }
+  }
+  __gl_meshDeleteMesh( mesh );
+  tess->polygonData= NULL;
+  tess->mesh = NULL;
+}
+
+
+/*XXXblythe unused function*/
+#if 0
+void GLAPIENTRY
+gluDeleteMesh( GLUmesh *mesh )
+{
+  __gl_meshDeleteMesh( mesh );
+}
+#endif
+
+
+
+/*******************************************************/
+
+/* Obsolete calls -- for backward compatibility */
+
+#if 0
+void GLAPIENTRY
+gluBeginPolygon( GLUtesselator *tess )
+{
+  gluTessBeginPolygon( tess, NULL );
+  gluTessBeginContour( tess );
+}
+
+
+/*ARGSUSED*/
+void GLAPIENTRY
+gluNextContour( GLUtesselator *tess, GLenum type )
+{
+  gluTessEndContour( tess );
+  gluTessBeginContour( tess );
+}
+
+
+void GLAPIENTRY
+gluEndPolygon( GLUtesselator *tess )
+{
+  gluTessEndContour( tess );
+  gluTessEndPolygon( tess );
+}
+#endif
diff --git a/third_party/glu/libtess/tess.h b/third_party/glu/libtess/tess.h
new file mode 100644
index 0000000..d8d3d21
--- /dev/null
+++ b/third_party/glu/libtess/tess.h
@@ -0,0 +1,172 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/tess.h#6 $
+*/
+
+#ifndef __tess_h_
+#define __tess_h_
+
+#include <sk_glu.h>
+#include <setjmp.h>
+#include "mesh.h"
+#include "dict.h"
+#include "priorityq.h"
+
+/* The begin/end calls must be properly nested.  We keep track of
+ * the current state to enforce the ordering.
+ */
+enum TessState { T_DORMANT, T_IN_POLYGON, T_IN_CONTOUR };
+
+/* We cache vertex data for single-contour polygons so that we can
+ * try a quick-and-dirty decomposition first.
+ */
+#define TESS_MAX_CACHE	100
+
+typedef struct CachedVertex {
+  GLdouble	coords[3];
+  void		*data;
+} CachedVertex;
+
+struct GLUtesselator {
+
+  /*** state needed for collecting the input data ***/
+
+  enum TessState state;		/* what begin/end calls have we seen? */
+
+  GLUhalfEdge	*lastEdge;	/* lastEdge->Org is the most recent vertex */
+  GLUmesh	*mesh;		/* stores the input contours, and eventually
+                                   the tessellation itself */
+
+  void		(GLAPIENTRY *callError)( GLenum errnum );
+
+  /*** state needed for projecting onto the sweep plane ***/
+
+  GLdouble	normal[3];	/* user-specified normal (if provided) */
+  GLdouble	sUnit[3];	/* unit vector in s-direction (debugging) */
+  GLdouble	tUnit[3];	/* unit vector in t-direction (debugging) */
+
+  /*** state needed for the line sweep ***/
+
+  GLdouble	relTolerance;	/* tolerance for merging features */
+  GLenum	windingRule;	/* rule for determining polygon interior */
+  GLboolean	fatalError;	/* fatal error: needed combine callback */
+
+  Dict		*dict;		/* edge dictionary for sweep line */
+  PriorityQ	*pq;		/* priority queue of vertex events */
+  GLUvertex	*event;		/* current sweep event being processed */
+
+  void		(GLAPIENTRY *callCombine)( GLdouble coords[3], void *data[4],
+			        GLfloat weight[4], void **outData );
+
+  /*** state needed for rendering callbacks (see render.c) ***/
+
+  GLboolean	flagBoundary;	/* mark boundary edges (use EdgeFlag) */
+  GLboolean	boundaryOnly;	/* Extract contours, not triangles */
+  GLUface	*lonelyTriList;
+    /* list of triangles which could not be rendered as strips or fans */
+
+  void		(GLAPIENTRY *callBegin)( GLenum type );
+  void		(GLAPIENTRY *callEdgeFlag)( GLboolean boundaryEdge );
+  void		(GLAPIENTRY *callVertex)( void *data );
+  void		(GLAPIENTRY *callEnd)( void );
+  void		(GLAPIENTRY *callMesh)( GLUmesh *mesh );
+
+
+  /*** state needed to cache single-contour polygons for renderCache() */
+
+  GLboolean	emptyCache;		/* empty cache on next vertex() call */
+  int		cacheCount;		/* number of cached vertices */
+  CachedVertex	cache[TESS_MAX_CACHE];	/* the vertex data */
+
+  /*** rendering callbacks that also pass polygon data  ***/ 
+  void		(GLAPIENTRY *callBeginData)( GLenum type, void *polygonData );
+  void		(GLAPIENTRY *callEdgeFlagData)( GLboolean boundaryEdge, 
+				     void *polygonData );
+  void		(GLAPIENTRY *callVertexData)( void *data, void *polygonData );
+  void		(GLAPIENTRY *callEndData)( void *polygonData );
+  void		(GLAPIENTRY *callErrorData)( GLenum errnum, void *polygonData );
+  void		(GLAPIENTRY *callCombineData)( GLdouble coords[3], void *data[4],
+				    GLfloat weight[4], void **outData,
+				    void *polygonData );
+
+  jmp_buf env;			/* place to jump to when memAllocs fail */
+
+  void *polygonData;		/* client data for current polygon */
+};
+
+void GLAPIENTRY __gl_noBeginData( GLenum type, void *polygonData );
+void GLAPIENTRY __gl_noEdgeFlagData( GLboolean boundaryEdge, void *polygonData );
+void GLAPIENTRY __gl_noVertexData( void *data, void *polygonData );
+void GLAPIENTRY __gl_noEndData( void *polygonData );
+void GLAPIENTRY __gl_noErrorData( GLenum errnum, void *polygonData );
+void GLAPIENTRY __gl_noCombineData( GLdouble coords[3], void *data[4],
+			 GLfloat weight[4], void **outData,
+			 void *polygonData );
+
+#define CALL_BEGIN_OR_BEGIN_DATA(a) \
+   if (tess->callBeginData != &__gl_noBeginData) \
+      (*tess->callBeginData)((a),tess->polygonData); \
+   else (*tess->callBegin)((a));
+
+#define CALL_VERTEX_OR_VERTEX_DATA(a) \
+   if (tess->callVertexData != &__gl_noVertexData) \
+      (*tess->callVertexData)((a),tess->polygonData); \
+   else (*tess->callVertex)((a));
+
+#define CALL_EDGE_FLAG_OR_EDGE_FLAG_DATA(a) \
+   if (tess->callEdgeFlagData != &__gl_noEdgeFlagData) \
+      (*tess->callEdgeFlagData)((a),tess->polygonData); \
+   else (*tess->callEdgeFlag)((a));
+
+#define CALL_END_OR_END_DATA() \
+   if (tess->callEndData != &__gl_noEndData) \
+      (*tess->callEndData)(tess->polygonData); \
+   else (*tess->callEnd)();
+
+#define CALL_COMBINE_OR_COMBINE_DATA(a,b,c,d) \
+   if (tess->callCombineData != &__gl_noCombineData) \
+      (*tess->callCombineData)((a),(b),(c),(d),tess->polygonData); \
+   else (*tess->callCombine)((a),(b),(c),(d));
+
+#define CALL_ERROR_OR_ERROR_DATA(a) \
+   if (tess->callErrorData != &__gl_noErrorData) \
+      (*tess->callErrorData)((a),tess->polygonData); \
+   else (*tess->callError)((a));
+
+#endif
diff --git a/third_party/glu/libtess/tessmono.c b/third_party/glu/libtess/tessmono.c
new file mode 100644
index 0000000..2e2cc79
--- /dev/null
+++ b/third_party/glu/libtess/tessmono.c
@@ -0,0 +1,208 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/tessmono.c#5 $
+*/
+
+#include "gluos.h"
+#include <stdlib.h>
+#include "geom.h"
+#include "mesh.h"
+#include "tessmono.h"
+#include <assert.h>
+
+#define AddWinding(eDst,eSrc)	(eDst->winding += eSrc->winding, \
+				 eDst->Sym->winding += eSrc->Sym->winding)
+
+/* __gl_meshTessellateMonoRegion( face ) tessellates a monotone region
+ * (what else would it do??)  The region must consist of a single
+ * loop of half-edges (see mesh.h) oriented CCW.  "Monotone" in this
+ * case means that any vertical line intersects the interior of the
+ * region in a single interval.  
+ *
+ * Tessellation consists of adding interior edges (actually pairs of
+ * half-edges), to split the region into non-overlapping triangles.
+ *
+ * The basic idea is explained in Preparata and Shamos (which I don''t
+ * have handy right now), although their implementation is more
+ * complicated than this one.  The are two edge chains, an upper chain
+ * and a lower chain.  We process all vertices from both chains in order,
+ * from right to left.
+ *
+ * The algorithm ensures that the following invariant holds after each
+ * vertex is processed: the untessellated region consists of two
+ * chains, where one chain (say the upper) is a single edge, and
+ * the other chain is concave.  The left vertex of the single edge
+ * is always to the left of all vertices in the concave chain.
+ *
+ * Each step consists of adding the rightmost unprocessed vertex to one
+ * of the two chains, and forming a fan of triangles from the rightmost
+ * of two chain endpoints.  Determining whether we can add each triangle
+ * to the fan is a simple orientation test.  By making the fan as large
+ * as possible, we restore the invariant (check it yourself).
+ */
+int __gl_meshTessellateMonoRegion( GLUface *face )
+{
+  GLUhalfEdge *up, *lo;
+
+  /* All edges are oriented CCW around the boundary of the region.
+   * First, find the half-edge whose origin vertex is rightmost.
+   * Since the sweep goes from left to right, face->anEdge should
+   * be close to the edge we want.
+   */
+  up = face->anEdge;
+  assert( up->Lnext != up && up->Lnext->Lnext != up );
+
+  for( ; VertLeq( up->Dst, up->Org ); up = up->Lprev )
+    ;
+  for( ; VertLeq( up->Org, up->Dst ); up = up->Lnext )
+    ;
+  lo = up->Lprev;
+
+  while( up->Lnext != lo ) {
+    if( VertLeq( up->Dst, lo->Org )) {
+      /* up->Dst is on the left.  It is safe to form triangles from lo->Org.
+       * The EdgeGoesLeft test guarantees progress even when some triangles
+       * are CW, given that the upper and lower chains are truly monotone.
+       */
+      while( lo->Lnext != up && (EdgeGoesLeft( lo->Lnext )
+	     || EdgeSign( lo->Org, lo->Dst, lo->Lnext->Dst ) <= 0 )) {
+	GLUhalfEdge *tempHalfEdge= __gl_meshConnect( lo->Lnext, lo );
+	if (tempHalfEdge == NULL) return 0;
+	lo = tempHalfEdge->Sym;
+      }
+      lo = lo->Lprev;
+    } else {
+      /* lo->Org is on the left.  We can make CCW triangles from up->Dst. */
+      while( lo->Lnext != up && (EdgeGoesRight( up->Lprev )
+	     || EdgeSign( up->Dst, up->Org, up->Lprev->Org ) >= 0 )) {
+	GLUhalfEdge *tempHalfEdge= __gl_meshConnect( up, up->Lprev );
+	if (tempHalfEdge == NULL) return 0;
+	up = tempHalfEdge->Sym;
+      }
+      up = up->Lnext;
+    }
+  }
+
+  /* Now lo->Org == up->Dst == the leftmost vertex.  The remaining region
+   * can be tessellated in a fan from this leftmost vertex.
+   */
+  assert( lo->Lnext != up );
+  while( lo->Lnext->Lnext != up ) {
+    GLUhalfEdge *tempHalfEdge= __gl_meshConnect( lo->Lnext, lo );
+    if (tempHalfEdge == NULL) return 0;
+    lo = tempHalfEdge->Sym;
+  }
+
+  return 1;
+}
+
+
+/* __gl_meshTessellateInterior( mesh ) tessellates each region of
+ * the mesh which is marked "inside" the polygon.  Each such region
+ * must be monotone.
+ */
+int __gl_meshTessellateInterior( GLUmesh *mesh )
+{
+  GLUface *f, *next;
+
+  /*LINTED*/
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = next ) {
+    /* Make sure we don''t try to tessellate the new triangles. */
+    next = f->next;
+    if( f->inside ) {
+      if ( !__gl_meshTessellateMonoRegion( f ) ) return 0;
+    }
+  }
+
+  return 1;
+}
+
+
+/* __gl_meshDiscardExterior( mesh ) zaps (ie. sets to NULL) all faces
+ * which are not marked "inside" the polygon.  Since further mesh operations
+ * on NULL faces are not allowed, the main purpose is to clean up the
+ * mesh so that exterior loops are not represented in the data structure.
+ */
+void __gl_meshDiscardExterior( GLUmesh *mesh )
+{
+  GLUface *f, *next;
+
+  /*LINTED*/
+  for( f = mesh->fHead.next; f != &mesh->fHead; f = next ) {
+    /* Since f will be destroyed, save its next pointer. */
+    next = f->next;
+    if( ! f->inside ) {
+      __gl_meshZapFace( f );
+    }
+  }
+}
+
+#define MARKED_FOR_DELETION	0x7fffffff
+
+/* __gl_meshSetWindingNumber( mesh, value, keepOnlyBoundary ) resets the
+ * winding numbers on all edges so that regions marked "inside" the
+ * polygon have a winding number of "value", and regions outside
+ * have a winding number of 0.
+ *
+ * If keepOnlyBoundary is TRUE, it also deletes all edges which do not
+ * separate an interior region from an exterior one.
+ */
+int __gl_meshSetWindingNumber( GLUmesh *mesh, int value,
+			        GLboolean keepOnlyBoundary )
+{
+  GLUhalfEdge *e, *eNext;
+
+  for( e = mesh->eHead.next; e != &mesh->eHead; e = eNext ) {
+    eNext = e->next;
+    if( e->Rface->inside != e->Lface->inside ) {
+
+      /* This is a boundary edge (one side is interior, one is exterior). */
+      e->winding = (e->Lface->inside) ? value : -value;
+    } else {
+
+      /* Both regions are interior, or both are exterior. */
+      if( ! keepOnlyBoundary ) {
+	e->winding = 0;
+      } else {
+	if ( !__gl_meshDelete( e ) ) return 0;
+      }
+    }
+  }
+  return 1;
+}
diff --git a/third_party/glu/libtess/tessmono.h b/third_party/glu/libtess/tessmono.h
new file mode 100644
index 0000000..6a71ee8
--- /dev/null
+++ b/third_party/glu/libtess/tessmono.h
@@ -0,0 +1,78 @@
+/*
+** License Applicability. Except to the extent portions of this file are
+** made subject to an alternative license as permitted in the SGI Free
+** Software License B, Version 1.1 (the "License"), the contents of this
+** file are subject only to the provisions of the License. You may not use
+** this file except in compliance with the License. You may obtain a copy
+** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
+** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
+** 
+** http://oss.sgi.com/projects/FreeB
+** 
+** Note that, as provided in the License, the Software is distributed on an
+** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
+** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
+** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
+** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+** 
+** Original Code. The Original Code is: OpenGL Sample Implementation,
+** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
+** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
+** Copyright in any portions created by third parties is as indicated
+** elsewhere herein. All Rights Reserved.
+** 
+** Additional Notice Provisions: The application programming interfaces
+** established by SGI in conjunction with the Original Code are The
+** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
+** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
+** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
+** Window System(R) (Version 1.3), released October 19, 1998. This software
+** was created using the OpenGL(R) version 1.2.1 Sample Implementation
+** published by SGI, but has not been independently verified as being
+** compliant with the OpenGL(R) version 1.2.1 Specification.
+**
+*/
+/*
+** Author: Eric Veach, July 1994.
+**
+** $Date$ $Revision$
+** $Header: //depot/main/gfx/lib/glu/libtess/tessmono.h#5 $
+*/
+
+#ifndef __tessmono_h_
+#define __tessmono_h_
+
+/* __gl_meshTessellateMonoRegion( face ) tessellates a monotone region
+ * (what else would it do??)  The region must consist of a single
+ * loop of half-edges (see mesh.h) oriented CCW.  "Monotone" in this
+ * case means that any vertical line intersects the interior of the
+ * region in a single interval.  
+ *
+ * Tessellation consists of adding interior edges (actually pairs of
+ * half-edges), to split the region into non-overlapping triangles.
+ *
+ * __gl_meshTessellateInterior( mesh ) tessellates each region of
+ * the mesh which is marked "inside" the polygon.  Each such region
+ * must be monotone.
+ *
+ * __gl_meshDiscardExterior( mesh ) zaps (ie. sets to NULL) all faces
+ * which are not marked "inside" the polygon.  Since further mesh operations
+ * on NULL faces are not allowed, the main purpose is to clean up the
+ * mesh so that exterior loops are not represented in the data structure.
+ *
+ * __gl_meshSetWindingNumber( mesh, value, keepOnlyBoundary ) resets the
+ * winding numbers on all edges so that regions marked "inside" the
+ * polygon have a winding number of "value", and regions outside
+ * have a winding number of 0.
+ *
+ * If keepOnlyBoundary is TRUE, it also deletes all edges which do not
+ * separate an interior region from an exterior one.
+ */
+
+int __gl_meshTessellateMonoRegion( GLUface *face );
+int __gl_meshTessellateInterior( GLUmesh *mesh );
+void __gl_meshDiscardExterior( GLUmesh *mesh );
+int __gl_meshSetWindingNumber( GLUmesh *mesh, int value,
+			        GLboolean keepOnlyBoundary );
+
+#endif
diff --git a/third_party/glu/sk_glu.h b/third_party/glu/sk_glu.h
new file mode 100644
index 0000000..7b2b261
--- /dev/null
+++ b/third_party/glu/sk_glu.h
@@ -0,0 +1,141 @@
+/*
+ * Copyright 2010, Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *     * Neither the name of Google Inc. nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+// This header is a skeleton substitute for GL/glu.h which contains
+// only the definitions, constants and function declarations necessary
+// to compile the GLU tessellator.
+
+#ifndef SK_GLU_H_
+#define SK_GLU_H_
+
+#include "gluos.h"
+
+/* Primitives */
+#define GL_LINE_LOOP                            0x0002
+#define GL_TRIANGLES                            0x0004
+#define GL_TRIANGLE_STRIP                       0x0005
+#define GL_TRIANGLE_FAN                         0x0006
+
+/* ErrorCode */
+#define GLU_INVALID_ENUM                     100900
+#define GLU_INVALID_VALUE                    100901
+#define GLU_OUT_OF_MEMORY                    100902
+
+/* TessCallback */
+#define GLU_TESS_BEGIN                       100100
+#define GLU_BEGIN                            100100
+#define GLU_TESS_VERTEX                      100101
+#define GLU_VERTEX                           100101
+#define GLU_TESS_END                         100102
+#define GLU_END                              100102
+#define GLU_TESS_ERROR                       100103
+#define GLU_TESS_EDGE_FLAG                   100104
+#define GLU_EDGE_FLAG                        100104
+#define GLU_TESS_COMBINE                     100105
+#define GLU_TESS_BEGIN_DATA                  100106
+#define GLU_TESS_VERTEX_DATA                 100107
+#define GLU_TESS_END_DATA                    100108
+#define GLU_TESS_ERROR_DATA                  100109
+#define GLU_TESS_EDGE_FLAG_DATA              100110
+#define GLU_TESS_COMBINE_DATA                100111
+
+/* TessContour */
+#define GLU_CW                               100120
+#define GLU_CCW                              100121
+#define GLU_INTERIOR                         100122
+#define GLU_EXTERIOR                         100123
+#define GLU_UNKNOWN                          100124
+
+/* TessProperty */
+#define GLU_TESS_WINDING_RULE                100140
+#define GLU_TESS_BOUNDARY_ONLY               100141
+#define GLU_TESS_TOLERANCE                   100142
+
+/* TessError */
+#define GLU_TESS_ERROR1                      100151
+#define GLU_TESS_ERROR2                      100152
+#define GLU_TESS_ERROR3                      100153
+#define GLU_TESS_ERROR4                      100154
+#define GLU_TESS_ERROR5                      100155
+#define GLU_TESS_ERROR6                      100156
+#define GLU_TESS_ERROR7                      100157
+#define GLU_TESS_ERROR8                      100158
+#define GLU_TESS_MISSING_BEGIN_POLYGON       100151
+#define GLU_TESS_MISSING_BEGIN_CONTOUR       100152
+#define GLU_TESS_MISSING_END_POLYGON         100153
+#define GLU_TESS_MISSING_END_CONTOUR         100154
+#define GLU_TESS_COORD_TOO_LARGE             100155
+#define GLU_TESS_NEED_COMBINE_CALLBACK       100156
+
+/* TessWinding */
+#define GLU_TESS_WINDING_ODD                 100130
+#define GLU_TESS_WINDING_NONZERO             100131
+#define GLU_TESS_WINDING_POSITIVE            100132
+#define GLU_TESS_WINDING_NEGATIVE            100133
+#define GLU_TESS_WINDING_ABS_GEQ_TWO         100134
+
+#define GLU_TESS_MAX_COORD 1.0e150
+
+typedef struct GLUtesselator GLUtesselator;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+extern GLUtesselator * GLAPIENTRY Sk_gluNewTess(void);
+extern void GLAPIENTRY Sk_gluDeleteTess(GLUtesselator *tess);
+extern void GLAPIENTRY Sk_gluTessProperty(GLUtesselator *tess,
+                                                GLenum which,
+                                                GLdouble value);
+extern void GLAPIENTRY Sk_gluGetTessProperty(GLUtesselator *tess,
+                                                   GLenum which,
+                                                   GLdouble *value);
+extern void GLAPIENTRY Sk_gluTessNormal(GLUtesselator *tess,
+                                              GLdouble x,
+                                              GLdouble y,
+                                              GLdouble z);
+extern void GLAPIENTRY Sk_gluTessCallback(GLUtesselator *tess,
+                                                GLenum which,
+                                                void (GLAPIENTRY *fn)());
+extern void GLAPIENTRY Sk_gluTessVertex(GLUtesselator *tess,
+                                              GLdouble coords[3],
+                                              void *data);
+extern void GLAPIENTRY Sk_gluTessBeginPolygon(GLUtesselator *tess,
+                                                    void *data);
+extern void GLAPIENTRY Sk_gluTessBeginContour(GLUtesselator *tess);
+extern void GLAPIENTRY Sk_gluTessEndContour(GLUtesselator *tess);
+extern void GLAPIENTRY Sk_gluTessEndPolygon(GLUtesselator *tess);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif  // SK_GLU_H_