Remove 'pv' parameter from Line/Tri/Point funcs. The provoking vertex

is always the last vertex parameter.
Modify clipping to preserve pv colors.
Modify swrast and X11 driver to expect the pv in the last vertex
(was looking in the first vertex previously).
Remove all handling of flatshading from swrast_setup.

Allow drivers to override the unclipped render tabs in tnl_render_stage
directly.  (Like in 3.4).  Removed fxsimplerender stage.

Modified t_vb_rendertmp.h to remove the need for 'parity' arguments
in RENDER_TRI macros.

9 files changed, 454 insertions, 667 deletions

diff --git a/src/mesa/tnl/t_array_import.c b/src/mesa/tnl/t_array_import.c
index fc2c3d1944..008cc23c64 100644
--- a/src/mesa/tnl/t_array_import.c
+++ b/src/mesa/tnl/t_array_import.c
@@ -1,4 +1,4 @@
-/* $Id: t_array_import.c,v 1.3 2000/12/28 22:11:05 keithw Exp $ */
+/* $Id: t_array_import.c,v 1.4 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -301,11 +301,6 @@ static void _tnl_upgrade_client_data( GLcontext *ctx,
 	    _tnl_import_texcoord( ctx, i, writeable, stride );
 	 }
    
-   if ((required & VERT_EDGE) && (VB->EdgeFlagPtr->flags & flags)) {
-      ASSERT(VB->EdgeFlagPtr == &inputs->EdgeFlag);
-      _tnl_import_edgeflag( ctx, writeable, stride );
-   }
-
    VB->importable_data &= ~required;
 }
 
@@ -386,11 +381,8 @@ void _tnl_vb_bind_arrays( GLcontext *ctx, GLint start, GLsizei count )
    }
     
    if (inputs & VERT_EDGE) {
-      if (imports & VERT_EDGE) {
-	 _tnl_import_edgeflag( ctx, 0, 0 ); 
-	 tmp->EdgeFlag.count = VB->Count;
-      }
-      VB->EdgeFlagPtr = &tmp->EdgeFlag;
+      _tnl_import_edgeflag( ctx, GL_TRUE, sizeof(GLboolean) ); 
+      VB->EdgeFlag = (GLboolean *) tmp->EdgeFlag.data;
    }
     
    if (inputs & VERT_SPEC_RGB) {
diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c
index 5b879d8a83..7d04210eb6 100644
--- a/src/mesa/tnl/t_context.c
+++ b/src/mesa/tnl/t_context.c
@@ -1,4 +1,4 @@
-/* $Id: t_context.c,v 1.7 2000/12/26 05:09:32 keithw Exp $ */
+/* $Id: t_context.c,v 1.8 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -69,6 +69,9 @@ install_driver_callbacks( GLcontext *ctx )
    ctx->Driver.MakeCurrent = _tnl_MakeCurrent;
    ctx->Driver.BeginCallList = _tnl_BeginCallList;
    ctx->Driver.EndCallList = _tnl_EndCallList;
+
+   ctx->Driver.RenderTabElts = _tnl_render_tab_elts;
+   ctx->Driver.RenderTabVerts = _tnl_render_tab_verts;
 }
 
 
@@ -111,7 +114,7 @@ _tnl_CreateContext( GLcontext *ctx )
    /* Set a few default values in the driver struct.
     */
    install_driver_callbacks(ctx);
-   ctx->Driver.NeedFlush = FLUSH_UPDATE_CURRENT|FLUSH_STORED_VERTICES;
+   ctx->Driver.NeedFlush = FLUSH_UPDATE_CURRENT;
    ctx->Driver.CurrentExecPrimitive = PRIM_OUTSIDE_BEGIN_END;
    ctx->Driver.CurrentSavePrimitive = PRIM_UNKNOWN;
 
diff --git a/src/mesa/tnl/t_context.h b/src/mesa/tnl/t_context.h
index 62d260681c..8e3952a412 100644
--- a/src/mesa/tnl/t_context.h
+++ b/src/mesa/tnl/t_context.h
@@ -1,4 +1,4 @@
-/* $Id: t_context.h,v 1.8 2000/12/27 22:30:29 keithw Exp $ */
+/* $Id: t_context.h,v 1.9 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -226,6 +226,12 @@ struct vertex_arrays
 
 typedef struct gl_material GLmaterial;
 
+typedef void (*interp_func)( GLcontext *ctx,
+			     GLfloat t, GLuint dst, GLuint in, GLuint out,
+			     GLboolean force_boundary );
+
+typedef void (*copy_pv_func)( GLcontext *ctx, GLuint dst, GLuint src );
+
 /* Contains the current state of a running pipeline.
  */
 typedef struct vertex_buffer
@@ -251,7 +257,7 @@ typedef struct vertex_buffer
    GLubyte     *ClipMask;		        /* VERT_CLIP (4) */
    GLvector3f  *NormalPtr;	                /* VERT_NORM */
    GLfloat     *NormalLengthPtr;                /* VERT_NORM (optional) */
-   GLvector1ub *EdgeFlagPtr;	                /* VERT_EDGE */
+   GLboolean   *EdgeFlag;	                /* VERT_EDGE */
    GLvector4f  *TexCoordPtr[MAX_TEXTURE_UNITS];	/* VERT_TEX_0..n */
    GLvector1ui *IndexPtr[2];	                /* VERT_INDEX */
    GLvector4ub *ColorPtr[2];	                /* VERT_RGBA */
@@ -276,9 +282,10 @@ typedef struct vertex_buffer
     */
 
    GLuint LastClipped;
-   void *interpfunc;
-   /* Two pieces of private data from _tnl_render_stage that have no
-    * business being in this struct.
+   interp_func interpfunc;
+   copy_pv_func copypvfunc;
+   /* Private data from _tnl_render_stage that has no business being
+    * in this struct.  
     */
 
 } TNLvertexbuffer;
diff --git a/src/mesa/tnl/t_imm_api.c b/src/mesa/tnl/t_imm_api.c
index 528f101661..aeca38dd4f 100644
--- a/src/mesa/tnl/t_imm_api.c
+++ b/src/mesa/tnl/t_imm_api.c
@@ -292,7 +292,7 @@ _tnl_end( GLcontext *ctx )
 	 IM->FlushElt = 0;
       }
       
-      ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES;
+/*        ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES; */
    }
 
    IM->BeginState = state;      
diff --git a/src/mesa/tnl/t_imm_exec.c b/src/mesa/tnl/t_imm_exec.c
index 1549fa53af..4bdce0d18d 100644
--- a/src/mesa/tnl/t_imm_exec.c
+++ b/src/mesa/tnl/t_imm_exec.c
@@ -1,4 +1,4 @@
-/* $Id: t_imm_exec.c,v 1.4 2000/12/28 22:11:05 keithw Exp $ */
+/* $Id: t_imm_exec.c,v 1.5 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -198,7 +198,7 @@ static void _tnl_vb_bind_immediate( GLcontext *ctx, struct immediate *IM )
    VB->NormalPtr = 0;
    VB->NormalLengthPtr = 0;
    VB->FogCoordPtr = 0;
-   VB->EdgeFlagPtr = 0;
+   VB->EdgeFlag = 0;
    VB->IndexPtr[0] = 0;
    VB->IndexPtr[1] = 0;
    VB->ColorPtr[0] = 0;
@@ -259,10 +259,7 @@ static void _tnl_vb_bind_immediate( GLcontext *ctx, struct immediate *IM )
    }
 
    if (inputs & VERT_EDGE) {
-      tmp->EdgeFlag.data = IM->EdgeFlag + start;
-      tmp->EdgeFlag.start = IM->EdgeFlag + start;
-      tmp->EdgeFlag.count = count;
-      VB->EdgeFlagPtr = &tmp->EdgeFlag;
+      VB->EdgeFlag = IM->EdgeFlag + start;
    }
 
    if (inputs & VERT_RGBA) {
@@ -451,8 +448,8 @@ void _tnl_execute_cassette( GLcontext *ctx, struct immediate *IM )
     */
    _tnl_copy_immediate_vertices( ctx, IM );
 
-/*     if (ctx->Driver.CurrentExecPrimitive == GL_POLYGON+1) */
-/*        ctx->Driver.NeedFlush &= ~FLUSH_STORED_VERTICES; */
+   if (ctx->Driver.CurrentExecPrimitive == GL_POLYGON+1)
+      ctx->Driver.NeedFlush &= ~FLUSH_STORED_VERTICES;
 }
 
 
diff --git a/src/mesa/tnl/t_pipeline.h b/src/mesa/tnl/t_pipeline.h
index 9d433bde0a..3f64eb8ed9 100644
--- a/src/mesa/tnl/t_pipeline.h
+++ b/src/mesa/tnl/t_pipeline.h
@@ -1,4 +1,4 @@
-/* $Id: t_pipeline.h,v 1.4 2000/12/26 05:09:33 keithw Exp $ */
+/* $Id: t_pipeline.h,v 1.5 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -62,4 +62,10 @@ extern const struct gl_pipeline_stage _tnl_render_stage;
  */
 extern const struct gl_pipeline_stage *_tnl_default_pipeline[];
 
+
+/* Convenience routines provided by t_vb_render.c:
+ */
+extern render_func _tnl_render_tab_elts[];
+extern render_func _tnl_render_tab_verts[];
+
 #endif
diff --git a/src/mesa/tnl/t_vb_cliptmp.h b/src/mesa/tnl/t_vb_cliptmp.h
index 1aeb014e15..5957f4960c 100644
--- a/src/mesa/tnl/t_vb_cliptmp.h
+++ b/src/mesa/tnl/t_vb_cliptmp.h
@@ -1,4 +1,4 @@
-/* $Id: t_vb_cliptmp.h,v 1.4 2000/12/28 22:11:05 keithw Exp $ */
+/* $Id: t_vb_cliptmp.h,v 1.5 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -28,340 +28,45 @@
  */
 
 
-#define INSIDE( J ) !NEGATIVE(J)
-#define OUTSIDE( J ) NEGATIVE(J)
+#define CLIP_DOTPROD(K, A, B, C, D) X(K)*A + Y(K)*B + Z(K)*C + W(K)*D
 
-
-
-
-static GLuint TAG(userclip_line)( GLcontext *ctx, 
-				  GLuint *i, GLuint *j,
-				  interp_func interp )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   GLfloat (*coord)[4] = VB->ClipPtr->data;
-   GLuint ii = *i;
-   GLuint jj = *j;
-   GLuint p;
-
-   for (p=0;p<MAX_CLIP_PLANES;p++) {
-      if (ctx->Transform.ClipEnabled[p]) {
-	 GLfloat a = ctx->Transform._ClipUserPlane[p][0];
-	 GLfloat b = ctx->Transform._ClipUserPlane[p][1];
-	 GLfloat c = ctx->Transform._ClipUserPlane[p][2];
-	 GLfloat d = ctx->Transform._ClipUserPlane[p][3];
-
-	 GLfloat dpI = d*W(ii) + c*Z(ii) + b*Y(ii) + a*X(ii);
-	 GLfloat dpJ = d*W(jj) + c*Z(jj) + b*Y(jj) + a*X(jj);
-
-         GLuint flagI = OUTSIDE( dpI );
-         GLuint flagJ = OUTSIDE( dpJ );
-
-	 if (flagI ^ flagJ) {
-	    if (flagJ) {
-	       GLfloat t = dpI / (dpI - dpJ);
-	       VB->ClipMask[jj] |= CLIP_USER_BIT;
-	       jj = interp( ctx, t, ii, jj, GL_FALSE );
-	       VB->ClipMask[jj] = 0;
-	    } else {
-	       GLfloat t = dpJ / (dpJ - dpI);
-	       VB->ClipMask[ii] |= CLIP_USER_BIT;
-	       ii = interp( ctx, t, jj, ii, GL_FALSE );
-	       VB->ClipMask[ii] = 0;
-	    }
-	 }
-	 else if (flagI) 
-	    return 0;
-      }
-   }
-
-   *i = ii;
-   *j = jj;
-   return 1;
-}
-
-
-static GLuint TAG(userclip_polygon)( GLcontext *ctx, 
-				     GLuint n, 
-				     GLuint vlist[],
-				     interp_func interp )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   GLfloat (*coord)[4] = VB->ClipPtr->data;
-   GLuint vlist2[MAX_CLIPPED_VERTICES];
-   GLuint *inlist = vlist, *outlist = vlist2;
-   GLubyte *clipmask = VB->ClipMask;
-   GLuint p;
-   
-#define CLIP_DOTPROD(xx) d*W(xx) + c*Z(xx) + b*Y(xx) + a*X(xx)
-   
-   /* Can be speeded up to if vertex_stage actually saves the
-    * UserClipMask, and if it is used in this loop (after computing a
-    * UserClipOrMask).
-    */
-   for (p=0;p<MAX_CLIP_PLANES;p++) {
-      if (ctx->Transform.ClipEnabled[p]) {
-	 register float a = ctx->Transform._ClipUserPlane[p][0];
-	 register float b = ctx->Transform._ClipUserPlane[p][1];
-	 register float c = ctx->Transform._ClipUserPlane[p][2];
-	 register float d = ctx->Transform._ClipUserPlane[p][3];
-
-	 /* initialize prev to be last in the input list */
-	 GLuint idxPrev = inlist[n-1];
-	 GLfloat dpPrev = CLIP_DOTPROD(idxPrev);
-	 GLuint outcount = 0;
-	 GLuint i;
-	 
-         for (i = 0 ; i < n ; i++) {	    
-	    GLuint idx = inlist[i];
-	    GLfloat dp = CLIP_DOTPROD(idx);
-
-	    if (!NEGATIVE(dpPrev)) {
-	       outlist[outcount++] = idxPrev;
-	       clipmask[idxPrev] &= ~CLIP_USER_BIT;
-	    } else {
-	       clipmask[idxPrev] |= CLIP_USER_BIT;
-	    }
-
-
-	    if (DIFFERENT_SIGNS(dp, dpPrev)) {
-	       GLuint newvert;
-	       if (NEGATIVE(dp)) {
-		  /* Going out of bounds.  Avoid division by zero as we
-		   * know dp != dpPrev from DIFFERENT_SIGNS, above.
-		   */
-		  GLfloat t = dp / (dp - dpPrev);
-		  newvert = interp( ctx, t, idx, idxPrev, GL_TRUE );
-	       } else {
-		  /* Coming back in.
-		   */
-		  GLfloat t = dpPrev / (dpPrev - dp);
-		  newvert = interp( ctx, t, idxPrev, idx, GL_FALSE );
-	       }
-	       clipmask[newvert] = 0;
-	       outlist[outcount++] = newvert;
-	    }
-	    
-	    idxPrev = idx;
-	    dpPrev = dp;
-	 }
-
-	 if (outcount < 3)
-	    return 0;
-
-	 {
-            GLuint *tmp;
-            tmp = inlist;
-            inlist = outlist;
-            outlist = tmp;
-            n = outcount;
-         }
-      } /* if */
-   } /* for p */
-
-   if (inlist!=vlist) {
-      GLuint i;
-      for (i = 0 ; i < n ; i++) 
-	 vlist[i] = inlist[i];
-   }
-
-   return n;
-}
-
-
-/* This now calls the user clip functions if required.
- */
-static void TAG(viewclip_line)( GLcontext *ctx,
-				GLuint i, GLuint j,
-				GLubyte mask )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   interp_func interp = (interp_func) VB->interpfunc;
-   GLfloat (*coord)[4] = VB->ClipPtr->data;
-   GLuint ii = i, jj = j;
-   GLuint vlist[2];
-   GLuint n;
-
-   VB->LastClipped = VB->FirstClipped;
-
-/*
- * We use 6 instances of this code to clip against the 6 planes.
- */
-#define GENERAL_CLIP					\
-   if (mask & PLANE) {					\
-      GLfloat dpI = CLIP_DOTPROD( ii );			\
-      GLfloat dpJ = CLIP_DOTPROD( jj );			\
-							\
-      if (DIFFERENT_SIGNS(dpI, dpJ)) {			\
-	 if (NEGATIVE(dpJ)) {				\
-	    GLfloat t = dpI / (dpI - dpJ);		\
-            VB->ClipMask[jj] |= PLANE;			\
-	    jj = interp( ctx, t, ii, jj, GL_FALSE );	\
-            VB->ClipMask[jj] = 0;			\
-	 } else {					\
-  	    GLfloat t = dpJ / (dpJ - dpI);		\
-            VB->ClipMask[ii] |= PLANE;			\
-	    ii = interp( ctx, t, jj, ii, GL_FALSE );	\
-            VB->ClipMask[ii] = 0;			\
-	 }						\
-      }							\
-      else if (NEGATIVE(dpI))				\
-	 return;					\
-  }
-
-#undef CLIP_DOTPROD
-#define PLANE CLIP_RIGHT_BIT
-#define CLIP_DOTPROD(K) (- X(K) + W(K))
-
-   GENERAL_CLIP
-
-#undef CLIP_DOTPROD
-#undef PLANE
-#define PLANE CLIP_LEFT_BIT
-#define CLIP_DOTPROD(K) (X(K) + W(K))
-
-   GENERAL_CLIP
-
-#undef CLIP_DOTPROD
-#undef PLANE
-#define PLANE CLIP_TOP_BIT
-#define CLIP_DOTPROD(K) (- Y(K) + W(K))
-
-   GENERAL_CLIP
-
-#undef CLIP_DOTPROD
-#undef PLANE
-#define PLANE CLIP_BOTTOM_BIT
-#define CLIP_DOTPROD(K) (Y(K) + W(K))
-
-   GENERAL_CLIP
-
-#undef CLIP_DOTPROD
-#undef PLANE
-#define PLANE CLIP_FAR_BIT
-#define CLIP_DOTPROD(K) (- Z(K) + W(K))
-
-   if (SIZE >= 3) { 
-      GENERAL_CLIP
-   }
-
-#undef CLIP_DOTPROD
-#undef PLANE
-#define PLANE CLIP_NEAR_BIT
-#define CLIP_DOTPROD(K) (Z(K) + W(K))
-
-   if (SIZE >=3 ) { 
-      GENERAL_CLIP
-   }
-
-#undef CLIP_DOTPROD
-#undef PLANE
-#undef GENERAL_CLIP
-
-
-   if (mask & CLIP_USER_BIT) {
-      if ( TAG(userclip_line)( ctx, &ii, &jj, interp ) == 0 )
-	 return;
-   }
-
-   vlist[0] = ii;
-   vlist[1] = jj;
-   n = 2;
-
-   /* If necessary, project new vertices.
-    */
-   {
-      GLuint i, j;
-      GLfloat (*proj)[4] = VB->ProjectedClipPtr->data;
-      GLuint start = VB->FirstClipped;
-
-      for (i = 0; i < n; i++) {
-	 j = vlist[i];
-	 if (j >= start) {
-	    if (SIZE == 4 && W(j) != 0.0F) {
-	       GLfloat wInv = 1.0F / W(j);
-	       proj[j][0] = X(j) * wInv;
-	       proj[j][1] = Y(j) * wInv;
-	       proj[j][2] = Z(j) * wInv;
-	       proj[j][3] = wInv;
-	    } else {
-	       proj[j][0] = X(j);
-	       proj[j][1] = Y(j);
-	       proj[j][2] = Z(j);
-	       proj[j][3] = W(j);
-	    }
-	 }
-      }
-   }
-
-   if (ctx->Driver.BuildProjectedVertices) 
-      ctx->Driver.BuildProjectedVertices(ctx, 
-					 VB->FirstClipped, 
-					 VB->LastClipped,
-					 ~0);
-   
-
-   /* Render the new line.
-    */
-   ctx->Driver.LineFunc( ctx, ii, jj, j );
-}
-
-/* We now clip polygon triangles individually.  This is necessary to
- * avoid artifacts dependent on where the boundary of the VB falls
- * within the polygon.  As a result, there is an upper bound on the
- * number of vertices which may be created, and the test against VB_SIZE
- * is redundant.  
- */
-static void TAG(viewclip_polygon)( GLcontext *ctx, 
-				   GLuint n, GLuint vlist[], GLuint pv,
-				   GLubyte mask )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   interp_func interp = (interp_func) VB->interpfunc;
-   GLfloat (*coord)[4] = VB->ClipPtr->data;
-   GLuint vlist2[MAX_CLIPPED_VERTICES];
-   GLuint *inlist = vlist, *outlist = vlist2;
-   GLuint i;
-   GLubyte *clipmask = VB->ClipMask;
-
-   VB->LastClipped = VB->FirstClipped;
-
-   if (mask & CLIP_ALL_BITS) {
-
-#define GENERAL_CLIP							\
+#define POLY_CLIP( PLANE, A, B, C, D )					\
+do {									\
    if (mask & PLANE) {							\
-      GLuint idxPrev = inlist[n-1];					\
-      GLfloat dpPrev = CLIP_DOTPROD(idxPrev);				\
+      GLuint idxPrev = inlist[0];					\
+      GLfloat dpPrev = CLIP_DOTPROD(idxPrev, A, B, C, D );		\
       GLuint outcount = 0;						\
       GLuint i;								\
-									\
-      mask &= ~PLANE;							\
-									\
-      for (i = 0; i < n; i++) {  					\
+    									\
+      inlist[n] = inlist[0]; /* prevent rotation of vertices */		\
+      for (i = 1; i <= n; i++) {					\
 	 GLuint idx = inlist[i];					\
-	 GLfloat dp = CLIP_DOTPROD(idx);				\
+	 GLfloat dp = CLIP_DOTPROD(idx, A, B, C, D );			\
 									\
+         clipmask[idxPrev] |= PLANE;					\
 	 if (!NEGATIVE(dpPrev)) {					\
 	    outlist[outcount++] = idxPrev;				\
  	    clipmask[idxPrev] &= ~PLANE;				\
 	 }								\
 									\
 	 if (DIFFERENT_SIGNS(dp, dpPrev)) {				\
-            GLuint newvert;						\
+            GLuint newvert = VB->LastClipped++;				\
+            VB->ClipMask[newvert] = 0;					\
+            outlist[outcount++] = newvert;				\
 	    if (NEGATIVE(dp)) {						\
 	       /* Going out of bounds.  Avoid division by zero as we	\
 		* know dp != dpPrev from DIFFERENT_SIGNS, above.	\
 		*/							\
 	       GLfloat t = dp / (dp - dpPrev);				\
-      	       newvert = interp( ctx, t, idx, idxPrev, GL_TRUE );	\
-	    } else {							\
+               LINTERP_SZ( t, coord, newvert, idx, idxPrev, SIZE );	\
+      	       interp( ctx, t, newvert, idx, idxPrev, GL_TRUE );	\
+	    } else {				\
 	       /* Coming back in.					\
 		*/							\
 	       GLfloat t = dpPrev / (dpPrev - dp);			\
-	       newvert = interp( ctx, t, idxPrev, idx, GL_FALSE );	\
+               LINTERP_SZ( t, coord, newvert, idxPrev, idx, SIZE );	\
+	       interp( ctx, t, newvert, idxPrev, idx, GL_FALSE );	\
 	    }								\
-            clipmask[newvert] = mask;					\
-            outlist[outcount++] = newvert; 				\
 	 }								\
 									\
 	 idxPrev = idx;							\
@@ -377,125 +82,184 @@ static void TAG(viewclip_polygon)( GLcontext *ctx,
 	 outlist = tmp;							\
 	 n = outcount;							\
       }									\
-   }
-
-
-#define PLANE CLIP_RIGHT_BIT
-#define CLIP_DOTPROD(K) (- X(K) + W(K))
-
-   GENERAL_CLIP
-
-#undef CLIP_DOTPROD
-#undef PLANE
-
-
-#define PLANE CLIP_LEFT_BIT
-#define CLIP_DOTPROD(K) (X(K) + W(K))
-
-   GENERAL_CLIP
+   }									\
+} while (0)
+
+
+#define LINE_CLIP(PLANE, A, B, C, D )				\
+do {								\
+   if (mask & PLANE) {						\
+      GLfloat dpI = CLIP_DOTPROD( ii, A, B, C, D );		\
+      GLfloat dpJ = CLIP_DOTPROD( jj, A, B, C, D );		\
+								\
+      if (DIFFERENT_SIGNS(dpI, dpJ)) {				\
+         GLuint newvert = VB->LastClipped++;			\
+         VB->ClipMask[newvert] = 0;				\
+	 if (NEGATIVE(dpJ)) {					\
+	    GLfloat t = dpI / (dpI - dpJ);			\
+            VB->ClipMask[jj] |= PLANE;				\
+            LINTERP_SZ( t, coord, newvert, ii, jj, SIZE );	\
+	    interp( ctx, t, newvert, ii, jj, GL_FALSE );	\
+            jj = newvert;					\
+	 } else {						\
+  	    GLfloat t = dpJ / (dpJ - dpI);			\
+            VB->ClipMask[ii] |= PLANE;				\
+            LINTERP_SZ( t, coord, newvert, jj, ii, SIZE );	\
+	    interp( ctx, t, newvert, jj, ii, GL_FALSE );	\
+            ii = newvert;					\
+	 }							\
+      }								\
+      else if (NEGATIVE(dpI))					\
+	 return;						\
+  }								\
+} while (0)
+
+
+/* Project if necessary.
+ */
+static void TAG(build_proj_verts)( GLcontext *ctx )
+{
+   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
+   GLfloat (*coord)[4] = VB->ClipPtr->data;
+   GLfloat (*proj)[4] = VB->ProjectedClipPtr->data;
+   GLuint last = VB->LastClipped;
+   GLuint i;
 
-#undef CLIP_DOTPROD
-#undef PLANE
+   for (i = VB->FirstClipped; i < last; i++) {
+      if (VB->ClipMask[i] == 0) {
+	 if (SIZE == 4 && W(i) != 0.0F) {
+	    GLfloat wInv = 1.0F / W(i);
+	    proj[i][0] = X(i) * wInv;
+	    proj[i][1] = Y(i) * wInv;
+	    proj[i][2] = Z(i) * wInv;
+	    proj[i][3] = wInv;
+	 } else {
+	    proj[i][0] = X(i);
+	    proj[i][1] = Y(i);
+	    proj[i][2] = Z(i);
+	    proj[i][3] = W(i);
+	 }
+      }
+   }
 
-#define PLANE CLIP_TOP_BIT
-#define CLIP_DOTPROD(K) (- Y(K) + W(K))
+   ctx->Driver.BuildProjectedVertices(ctx, 
+				      VB->FirstClipped, 
+				      VB->LastClipped,
+				      ~0);
+}
 
-   GENERAL_CLIP
+/* Clip a line against the viewport and user clip planes.
+ */
+static void TAG(clip_line)( GLcontext *ctx,
+			    GLuint i, GLuint j,
+			    GLubyte mask )
+{
+   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
+   interp_func interp = (interp_func) VB->interpfunc;
+   GLfloat (*coord)[4] = VB->ClipPtr->data;
+   GLuint ii = i, jj = j, p;
 
-#undef CLIP_DOTPROD
-#undef PLANE
+   VB->LastClipped = VB->FirstClipped;
 
-#define PLANE CLIP_BOTTOM_BIT
-#define CLIP_DOTPROD(K) (Y(K) + W(K))
+   if (mask & 0x3f) {
+      LINE_CLIP( CLIP_RIGHT_BIT,  -1,  0,  0, 1 );
+      LINE_CLIP( CLIP_LEFT_BIT,    1,  0,  0, 1 );
+      LINE_CLIP( CLIP_TOP_BIT,     0, -1,  0, 1 );
+      LINE_CLIP( CLIP_BOTTOM_BIT,  0,  1,  0, 1 );
+      LINE_CLIP( CLIP_FAR_BIT,     0,  0, -1, 1 );
+      LINE_CLIP( CLIP_NEAR_BIT,    0,  0,  1, 1 );
+   }
 
-   GENERAL_CLIP
+   if (mask & CLIP_USER_BIT) {
+      for (p=0;p<MAX_CLIP_PLANES;p++) {
+	 if (ctx->Transform.ClipEnabled[p]) {
+	    GLfloat a = ctx->Transform._ClipUserPlane[p][0];
+	    GLfloat b = ctx->Transform._ClipUserPlane[p][1];
+	    GLfloat c = ctx->Transform._ClipUserPlane[p][2];
+	    GLfloat d = ctx->Transform._ClipUserPlane[p][3];
+	    LINE_CLIP( CLIP_USER_BIT, a, b, c, d );
+	 }
+      }
+   }
 
-#undef CLIP_DOTPROD
-#undef PLANE
+   TAG(build_proj_verts)( ctx );
 
-#define PLANE CLIP_FAR_BIT
-#define CLIP_DOTPROD(K) (- Z(K) + W(K))
+   if ((ctx->_TriangleCaps & DD_FLATSHADE) && j != jj)
+      VB->copypvfunc( ctx, jj, j );
 
-   if (SIZE >= 3) { 
-      GENERAL_CLIP
-   }
+   /* Render the new line.
+    */
+   ctx->Driver.LineFunc( ctx, ii, jj );
 
-#undef CLIP_DOTPROD
-#undef PLANE
+}
 
-#define PLANE CLIP_NEAR_BIT
-#define CLIP_DOTPROD(K) (Z(K) + W(K))
 
-   if (SIZE >=3 ) { 
-      GENERAL_CLIP
-   }
+/* Clip a triangle or quad against the viewport and user clip planes.
+ */
+static void TAG(clip_polygon)( GLcontext *ctx, 
+			       GLuint n, GLuint vlist[], 
+			       GLubyte mask )
+{
+   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
+   interp_func interp = (interp_func) VB->interpfunc;
+   GLfloat (*coord)[4] = VB->ClipPtr->data;
+   GLuint pv = vlist[0];
+   GLuint vlist2[MAX_CLIPPED_VERTICES];
+   GLuint *inlist = vlist, *outlist = vlist2;
+   GLuint p;
+   GLubyte *clipmask = VB->ClipMask;
 
-#undef CLIP_DOTPROD
-#undef PLANE
-#undef GENERAL_CLIP
+   VB->LastClipped = VB->FirstClipped;
 
-      if (inlist != vlist) 
-	 for (i = 0 ; i < n ; i++)
-	    vlist[i] = inlist[i];
+   if (mask & 0x3f) {
+      POLY_CLIP( CLIP_RIGHT_BIT,  -1,  0,  0, 1 );
+      POLY_CLIP( CLIP_LEFT_BIT,    1,  0,  0, 1 );
+      POLY_CLIP( CLIP_TOP_BIT,     0, -1,  0, 1 );
+      POLY_CLIP( CLIP_BOTTOM_BIT,  0,  1,  0, 1 );
+      POLY_CLIP( CLIP_FAR_BIT,     0,  0, -1, 1 );
+      POLY_CLIP( CLIP_NEAR_BIT,    0,  0,  1, 1 );
    }
 
-   /* Clip against user clipping planes in clip space. 
-    */
    if (mask & CLIP_USER_BIT) {
-      n = TAG(userclip_polygon)( ctx, n, vlist, interp );
-      if (n < 3) return;
+      for (p=0;p<MAX_CLIP_PLANES;p++) {
+	 if (ctx->Transform.ClipEnabled[p]) {
+	    GLfloat a = ctx->Transform._ClipUserPlane[p][0];
+	    GLfloat b = ctx->Transform._ClipUserPlane[p][1];
+	    GLfloat c = ctx->Transform._ClipUserPlane[p][2];
+	    GLfloat d = ctx->Transform._ClipUserPlane[p][3];
+	    POLY_CLIP( CLIP_USER_BIT, a, b, c, d );
+	 }
+      }
    }
 
-   /* Project if necessary.
-    */
-   {
-      GLuint i;
-      GLfloat (*proj)[4] = VB->ProjectedClipPtr->data;
-      GLuint first = VB->FirstClipped;
-
-      for (i = 0; i < n; i++) {
-	 GLuint j = vlist[i];
-	 if (j >= first) {
-	    if (SIZE == 4 && W(j) != 0.0F) {
-	       GLfloat wInv = 1.0F / W(j);
-	       proj[j][0] = X(j) * wInv;
-	       proj[j][1] = Y(j) * wInv;
-	       proj[j][2] = Z(j) * wInv;
-	       proj[j][3] = wInv;
-	    } else {
-	       proj[j][0] = X(j);
-	       proj[j][1] = Y(j);
-	       proj[j][2] = Z(j);
-	       proj[j][3] = W(j);
-	    }
-	 }
+   if (ctx->_TriangleCaps & DD_FLATSHADE) {
+      if (pv != inlist[0]) {
+	 ASSERT( inlist[0] >= VB->FirstClipped );
+	 VB->copypvfunc( ctx, inlist[0], pv );
       }
    }
 
-   if (ctx->Driver.BuildProjectedVertices)
-      ctx->Driver.BuildProjectedVertices(ctx, 
-					 VB->FirstClipped, 
-					 VB->LastClipped,
-					 ~0);
+   TAG(build_proj_verts)( ctx );
 
    /* Render the new vertices as an unclipped polygon. 
-    * Argh - need to pass in pv...
     */
    {
       GLuint *tmp = VB->Elts;
-      VB->Elts = vlist;
-      render_poly_pv_raw_elts( ctx, 0, n, PRIM_BEGIN|PRIM_END, pv );
+      VB->Elts = inlist;
+      ctx->Driver.RenderTabElts[GL_POLYGON]( ctx, 0, n, PRIM_BEGIN|PRIM_END );
       VB->Elts = tmp;
    }
 }
 
 
 
+
+
 #undef W
 #undef Z
 #undef Y
 #undef X
 #undef SIZE
 #undef TAG
-#undef INSIDE
-#undef OUTSIDE
+#undef POLY_CLIP
+#undef LINE_CLIP
diff --git a/src/mesa/tnl/t_vb_render.c b/src/mesa/tnl/t_vb_render.c
index b3fef23501..2616d294d3 100644
--- a/src/mesa/tnl/t_vb_render.c
+++ b/src/mesa/tnl/t_vb_render.c
@@ -1,4 +1,4 @@
-/* $Id: t_vb_render.c,v 1.5 2001/01/03 15:59:31 brianp Exp $ */
+/* $Id: t_vb_render.c,v 1.6 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -22,6 +22,9 @@
  * BRIAN PAUL 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.
+ *
+ * Author:
+ *    Keith Whitwell <keithw@valinux.com>
  */
 
 
@@ -33,11 +36,16 @@
  * and triangle rasterizers via the function pointers:
  *
  *    context->Driver.BuildProjectedVertices()
+ *
  *    context->Driver.PointsFunc()
  *    context->Driver.LineFunc()
  *    context->Driver.TriangleFunc()
  *    context->Driver.QuadFunc() 
  *
+ *    context->Driver.RenderTabVerts[] 
+ *    context->Driver.RenderTabElts[]
+ *
+ * None of these may be null.
  */
 
 
@@ -55,9 +63,6 @@
 #include "t_pipeline.h"
 
 
-typedef GLuint (*interp_func)( GLcontext *ctx,
-			       GLfloat t, GLuint in, GLuint out,
-			       GLboolean force_boundary );
 
 typedef void (*clip_line_func)( GLcontext *ctx,
 				GLuint i, GLuint j,
@@ -65,13 +70,8 @@ typedef void (*clip_line_func)( GLcontext *ctx,
 
 typedef void (*clip_poly_func)( GLcontext *ctx,
 				GLuint n, GLuint vlist[],
-				GLuint pv, GLubyte mask );
-
+				GLubyte mask );
 
-typedef void (*render_func)( GLcontext *ctx, 
-			     GLuint start,
-			     GLuint count,
-			     GLuint flags );
 
 
 
@@ -79,33 +79,19 @@ struct render_stage_data {
 
    /* Clipping functions for current state.
     */
-   interp_func interp; /* Clip interpolation function */
-   GLuint _ClipInputs;		     /* Inputs referenced by interpfunc */
-
+   interp_func interp;		/* Clip interpolation function */
+   copy_pv_func copypv;		/* Flatshade fixup function */
+   GLuint _ClipInputs;		/* Inputs referenced by interpfunc */
 };
 
 #define RENDER_STAGE_DATA(stage) ((struct render_stage_data *)stage->private)
 
-static void render_poly_pv_raw_elts( GLcontext *ctx, 
-				     GLuint start,
-				     GLuint count,
-				     GLuint flags,
-				     GLuint pv );
 				  
 /**********************************************************************/
 /*           Interpolate between pairs of vertices                    */
 /**********************************************************************/
 
 
-#define INTERP_RGBA    0x1
-#define INTERP_TEX     0x2
-#define INTERP_INDEX   0x4
-#define INTERP_SPEC    0x8
-#define INTERP_FOG     0x10
-#define INTERP_EDGE    0x20
-#define MAX_INTERP     0x40
-
-
 #define LINTERP_SZ( t, vec, to, a, b, sz )			\
 do {								\
    switch (sz) {						\
@@ -146,8 +132,50 @@ do {								\
 
 
 
-static interp_func interp_tab[0x80];
 
+#define INTERP_RGBA    0x1
+#define INTERP_TEX     0x2
+#define INTERP_INDEX   0x4
+#define INTERP_SPEC    0x8
+#define INTERP_FOG     0x10
+#define INTERP_EDGE    0x20
+#define MAX_INTERP     0x40
+
+static interp_func interp_tab[MAX_INTERP];
+static copy_pv_func copy_tab[MAX_INTERP];
+
+
+#define IND (0)
+#define NAME interp_none
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_FOG)
+#define NAME interp_FOG
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_TEX)
+#define NAME interp_TEX
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_FOG|INTERP_TEX)
+#define NAME interp_FOG_TEX
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_EDGE)
+#define NAME interp_EDGE
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_FOG|INTERP_EDGE)
+#define NAME interp_FOG_EDGE
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_TEX|INTERP_EDGE)
+#define NAME interp_TEX_EDGE
+#include "t_vb_interptmp.h"
+
+#define IND (INTERP_FOG|INTERP_TEX|INTERP_EDGE)
+#define NAME interp_FOG_TEX_EDGE
+#include "t_vb_interptmp.h"
 
 #define IND (INTERP_RGBA)
 #define NAME interp_RGBA
@@ -246,15 +274,34 @@ static interp_func interp_tab[0x80];
 #include "t_vb_interptmp.h"
 
 
+#define IND (INTERP_RGBA)
+#define NAME copy_RGBA
+#include "t_vb_flattmp.h"
+
+#define IND (INTERP_RGBA|INTERP_SPEC)
+#define NAME copy_RGBA_SPEC
+#include "t_vb_flattmp.h"
+
+#define IND (INTERP_INDEX)
+#define NAME copy_INDEX
+#include "t_vb_flattmp.h"
+
+
+
 
-static GLuint interp_invalid( GLcontext *ctx,
-			      GLfloat t, 
-			      GLuint in, GLuint out,
-			      GLboolean boundary )
+static void interp_invalid( GLcontext *ctx,
+			    GLfloat t, 
+			    GLuint dst, GLuint in, GLuint out,
+			    GLboolean boundary )
 {
    (void)(ctx && t && in && out && boundary);
    fprintf(stderr, "Invalid interpolation function in t_vbrender.c\n");
-   return in;
+}
+
+static void copy_invalid( GLcontext *ctx, GLuint dst, GLuint src )
+{
+   (void)(ctx && dst && src);
+   fprintf(stderr, "Invalid copy function in t_vbrender.c\n");
 }
 
 
@@ -266,8 +313,19 @@ static void interp_init( void )
     * the non-implemented combinations are reachable, but this gives
     * some safety from crashes.
     */
-   for (i = 0 ; i < Elements(interp_tab) ; i++)
+   for (i = 0 ; i < Elements(interp_tab) ; i++) {
       interp_tab[i] = interp_invalid;
+      copy_tab[i] = copy_invalid;
+   }
+
+   interp_tab[0] = interp_none;
+   interp_tab[INTERP_FOG] = interp_FOG;
+   interp_tab[INTERP_TEX] = interp_TEX;
+   interp_tab[INTERP_FOG|INTERP_TEX] = interp_FOG_TEX;
+   interp_tab[INTERP_EDGE] = interp_EDGE;
+   interp_tab[INTERP_FOG|INTERP_EDGE] = interp_FOG_EDGE;
+   interp_tab[INTERP_TEX|INTERP_EDGE] = interp_TEX_EDGE;
+   interp_tab[INTERP_FOG|INTERP_TEX|INTERP_EDGE] = interp_FOG_TEX_EDGE;
 
    interp_tab[INTERP_RGBA] = interp_RGBA;
    interp_tab[INTERP_RGBA|INTERP_SPEC] = interp_RGBA_SPEC;
@@ -276,7 +334,8 @@ static void interp_init( void )
    interp_tab[INTERP_RGBA|INTERP_TEX] = interp_RGBA_TEX;
    interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_TEX] = interp_RGBA_SPEC_TEX;
    interp_tab[INTERP_RGBA|INTERP_FOG|INTERP_TEX] = interp_RGBA_FOG_TEX;
-   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_FOG|INTERP_TEX] = interp_RGBA_SPEC_FOG_TEX;
+   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_FOG|INTERP_TEX] = 
+      interp_RGBA_SPEC_FOG_TEX;
    interp_tab[INTERP_INDEX] = interp_INDEX;
    interp_tab[INTERP_FOG|INTERP_INDEX] = interp_FOG_INDEX;
    interp_tab[INTERP_TEX|INTERP_INDEX] = interp_TEX_INDEX;
@@ -284,15 +343,26 @@ static void interp_init( void )
    interp_tab[INTERP_RGBA|INTERP_EDGE] = interp_RGBA_EDGE;
    interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_EDGE] = interp_RGBA_SPEC_EDGE;
    interp_tab[INTERP_RGBA|INTERP_FOG|INTERP_EDGE] = interp_RGBA_FOG_EDGE;
-   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_FOG|INTERP_EDGE] = interp_RGBA_SPEC_FOG_EDGE;
+   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_FOG|INTERP_EDGE] = 
+      interp_RGBA_SPEC_FOG_EDGE;
    interp_tab[INTERP_RGBA|INTERP_TEX|INTERP_EDGE] = interp_RGBA_TEX_EDGE;
-   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_TEX|INTERP_EDGE] = interp_RGBA_SPEC_TEX_EDGE;
-   interp_tab[INTERP_RGBA|INTERP_FOG|INTERP_TEX|INTERP_EDGE] = interp_RGBA_FOG_TEX_EDGE;
-   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_FOG|INTERP_TEX|INTERP_EDGE] = interp_RGBA_SPEC_FOG_TEX_EDGE;
+   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_TEX|INTERP_EDGE] = 
+      interp_RGBA_SPEC_TEX_EDGE;
+   interp_tab[INTERP_RGBA|INTERP_FOG|INTERP_TEX|INTERP_EDGE] = 
+      interp_RGBA_FOG_TEX_EDGE;
+   interp_tab[INTERP_RGBA|INTERP_SPEC|INTERP_FOG|INTERP_TEX|INTERP_EDGE] = 
+      interp_RGBA_SPEC_FOG_TEX_EDGE;
    interp_tab[INTERP_INDEX|INTERP_EDGE] = interp_INDEX_EDGE;
    interp_tab[INTERP_FOG|INTERP_INDEX|INTERP_EDGE] = interp_FOG_INDEX_EDGE;
    interp_tab[INTERP_TEX|INTERP_INDEX|INTERP_EDGE] = interp_TEX_INDEX_EDGE;
-   interp_tab[INTERP_FOG|INTERP_TEX|INTERP_INDEX|INTERP_EDGE] = interp_FOG_TEX_INDEX_EDGE;
+   interp_tab[INTERP_FOG|INTERP_TEX|INTERP_INDEX|INTERP_EDGE] = 
+      interp_FOG_TEX_INDEX_EDGE;
+
+
+   copy_tab[INTERP_RGBA] = copy_RGBA;
+   copy_tab[INTERP_RGBA|INTERP_SPEC] = copy_RGBA_SPEC;
+   copy_tab[INTERP_INDEX] = copy_INDEX;
+
 }
 
 
@@ -301,12 +371,15 @@ static void interp_init( void )
 /**********************************************************************/
 
 
-#if 0
-#define NEGATIVE(x) ((*(int *)&x)<0)
-#define DIFFERENT_SIGNS(a,b) ((a*b) < 0)
+#if defined(USE_IEEE) 
+#define NEGATIVE(x) ((*(GLuint *)&x) & (1<<31))
+#define DIFFERENT_SIGNS(x,y) (((*(GLuint *)&x)^(*(GLuint *)&y)) & (1<<31))
 #else
 #define NEGATIVE(x) (x < 0)
-#define DIFFERENT_SIGNS(a,b) ((a*b) < 0)
+#define DIFFERENT_SIGNS(x,y) (x * y <= 0 && x - y != 0)
+/* Could just use (x*y<0) except for the flatshading requirements.
+ * Maybe there's a better way?
+ */
 #endif
 
 #define W(i) coord[i][3]
@@ -336,176 +409,151 @@ static void interp_init( void )
 static clip_poly_func clip_poly_tab[5] = {
    0,
    0,
-   viewclip_polygon_2,
-   viewclip_polygon_3,
-   viewclip_polygon_4
+   clip_polygon_2,
+   clip_polygon_3,
+   clip_polygon_4
 };
 
 static clip_line_func clip_line_tab[5] = {
    0,
    0,
-   viewclip_line_2,
-   viewclip_line_3,
-   viewclip_line_4
+   clip_line_2,
+   clip_line_3,
+   clip_line_4
 };
 
 
-
 /**********************************************************************/
-/*                 Clip and render single primitives                  */
-/**********************************************************************/
-
-
-
-static INLINE void draw_line(GLcontext *ctx, GLuint v1, GLuint v2 )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   GLubyte c1 = VB->ClipMask[v1], c2 = VB->ClipMask[v2];
-   GLubyte ormask = c1|c2; 
-   if (!ormask) 
-      ctx->Driver.LineFunc( ctx, v1, v2, v2 );
-   else if (!(c1 & c2 & 0x3f))
-      clip_line_tab[VB->ClipPtr->size]( ctx, v1, v2, ormask );
-}
-
-static INLINE void draw_triangle(GLcontext *ctx, 
-				 GLuint v1, GLuint v2, GLuint v3,
-				 GLuint pv )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   GLubyte c1 = VB->ClipMask[v1], c2 = VB->ClipMask[v2], c3 = VB->ClipMask[v3];
-   GLubyte ormask = c1|c2|c3; 
-   if (!ormask) 
-      ctx->Driver.TriangleFunc( ctx, v1, v2, v3, pv );
-   else if (!(c1 & c2 & c3 & 0x3f)) {
-      GLuint vlist[MAX_CLIPPED_VERTICES];
-      ASSIGN_3V(vlist, v1, v2, v3 );
-      clip_poly_tab[VB->ClipPtr->size]( ctx, 3, vlist, pv, ormask );
-   }
-}
-
-
-static INLINE void draw_quad( GLcontext *ctx, 
-			      GLuint v1, GLuint v2, GLuint v3,
-			      GLuint v4, GLuint pv )
-{
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;   
-   GLubyte c1 = VB->ClipMask[v1], c2 = VB->ClipMask[v2];
-   GLubyte c3 = VB->ClipMask[v3], c4 = VB->ClipMask[v4];
-   GLubyte ormask = c1|c2|c3|c4; 
-   if (!ormask) 
-      ctx->Driver.QuadFunc( ctx, v1, v2, v3, v4, pv );
-   else if (!(c1 & c2 & c3 & c4 & 0x3f)) {
-      GLuint vlist[MAX_CLIPPED_VERTICES];
-      ASSIGN_4V(vlist, v1, v2, v3, v4 );
-      clip_poly_tab[VB->ClipPtr->size]( ctx, 4, vlist, pv, ormask );
-   }
-}
-
-
-/**********************************************************************/
-/*            Clip and render whole begin/end objects                 */
+/*              Clip and render whole begin/end objects               */
 /**********************************************************************/
 
 #define NEED_EDGEFLAG_SETUP (ctx->_TriangleCaps & DD_TRI_UNFILLED)
-#define EDGEFLAG_GET(idx) VB->EdgeFlagPtr->data[idx]
-#define EDGEFLAG_SET(idx, val) VB->EdgeFlagPtr->data[idx] = val
+#define EDGEFLAG_GET(idx) VB->EdgeFlag[idx]
+#define EDGEFLAG_SET(idx, val) VB->EdgeFlag[idx] = val
 
 
-/* Vertices, no clipping.
+/* Vertices, with the possibility of clipping.
  */
 #define RENDER_POINTS( start, count ) \
-   ctx->Driver.PointsFunc( ctx, start, count-1 )
+   ctx->Driver.PointsFunc( ctx, start, count )
 
-#define RENDER_LINE( i1, i ) \
-   ctx->Driver.LineFunc( ctx, i1, i, i )
+#define RENDER_LINE( v1, v2 )			\
+do {						\
+   GLubyte c1 = mask[v1], c2 = mask[v2];	\
+   GLubyte ormask = c1|c2;			\
+   if (!ormask)					\
+      LineFunc( ctx, v1, v2 );			\
+   else if (!(c1 & c2 & 0x3f))			\
+      clip_line_tab[sz]( ctx, v1, v2, ormask );	\
+} while (0)
 
-#define RENDER_TRI( i2, i1, i, pv, parity )		\
+#define RENDER_TRI( v1, v2, v3 )			\
 do {							\
-   if (parity)						\
-      ctx->Driver.TriangleFunc( ctx, i1, i2, i, pv );	\
-   else							\
-      ctx->Driver.TriangleFunc( ctx, i2, i1, i, pv );	\
+   GLubyte c1 = mask[v1], c2 = mask[v2], c3 = mask[v3];	\
+   GLubyte ormask = c1|c2|c3;				\
+   if (!ormask)						\
+      TriangleFunc( ctx, v1, v2, v3 );			\
+   else if (!(c1 & c2 & c3 & 0x3f)) {			\
+      GLuint vlist[MAX_CLIPPED_VERTICES];		\
+      ASSIGN_3V(vlist, v3, v1, v2 );			\
+      clip_poly_tab[sz]( ctx, 3, vlist, ormask );	\
+   }							\
 } while (0)
 
-#define RENDER_QUAD( i3, i2, i1, i, pv )	\
-   ctx->Driver.QuadFunc( ctx, i3, i2, i1, i, pv );
-
-#define TAG(x) x##_raw
+#define RENDER_QUAD( v1, v2, v3, v4 )			\
+do {							\
+   GLubyte c1 = mask[v1], c2 = mask[v2];		\
+   GLubyte c3 = mask[v3], c4 = mask[v4];		\
+   GLubyte ormask = c1|c2|c3|c4;			\
+   if (!ormask)						\
+      QuadFunc( ctx, v1, v2, v3, v4 );			\
+   else if (!(c1 & c2 & c3 & c4 & 0x3f)) {		\
+      GLuint vlist[MAX_CLIPPED_VERTICES];		\
+      ASSIGN_4V(vlist, v4, v1, v2, v3 );		\
+      clip_poly_tab[sz]( ctx, 4, vlist, ormask );	\
+   }							\
+} while (0)
 
-#define LOCAL_VARS					\
-    struct vertex_buffer *VB = &(TNL_CONTEXT(ctx)->vb);	\
-    (void) VB;
 
+#define LOCAL_VARS							\
+    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;			\
+    const GLuint * const elt = VB->Elts;				\
+    const GLubyte *mask = VB->ClipMask;					\
+    const GLuint sz = VB->ClipPtr->size;				\
+    const line_func LineFunc = ctx->Driver.LineFunc;			\
+    const triangle_func TriangleFunc = ctx->Driver.TriangleFunc;	\
+    const quad_func QuadFunc = ctx->Driver.QuadFunc;			\
+    (void) (LineFunc && TriangleFunc && QuadFunc);			\
+    (void) elt; (void) mask; (void) sz;
+
+#define TAG(x) clip_##x##_verts
 #define RESET_STIPPLE ctx->Driver.ResetLineStipple( ctx )
 #define RESET_OCCLUSION ctx->OcclusionResult = GL_TRUE;
 #define PRESERVE_VB_DEFS
 #include "t_vb_rendertmp.h"
 
 
-/* Elts, no clipping.
+
+/* Elts, with the possibility of clipping.
  */
 #undef ELT
 #undef TAG
-#undef LOCAL_VARS
-#define TAG(x) x##_raw_elts
 #define ELT(x) elt[x]
-#define LOCAL_VARS  				\
-    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb; \
-    const GLuint * const elt = VB->Elts;	\
-    (void) elt;
+#define TAG(x) clip_##x##_elts
 #include "t_vb_rendertmp.h"
 
 
+/**********************************************************************/
+/*                  Render whole begin/end objects                    */
+/**********************************************************************/
 
+#define NEED_EDGEFLAG_SETUP (ctx->_TriangleCaps & DD_TRI_UNFILLED)
+#define EDGEFLAG_GET(idx) VB->EdgeFlag[idx]
+#define EDGEFLAG_SET(idx, val) VB->EdgeFlag[idx] = val
 
-/* Vertices, with the possibility of clipping.
+
+/* Vertices, no clipping.
  */
-#define RENDER_POINTS( start, count )			\
-   ctx->Driver.PointsFunc( ctx, start, count-1 )
+#define RENDER_POINTS( start, count ) \
+   ctx->Driver.PointsFunc( ctx, start, count )
 
-#define RENDER_LINE( i1, i )			\
-   draw_line( ctx, i1, i )
+#define RENDER_LINE( v1, v2 ) \
+   LineFunc( ctx, v1, v2 )
 
-#define RENDER_TRI( i2, i1, i, pv, parity)	\
-do {						\
-  GLuint e2=i2, e1=i1;				\
-  if (parity) { GLuint t=e2; e2=e1; e1=t; }	\
-  draw_triangle(ctx,e2,e1,i,pv);	\
-} while (0)
+#define RENDER_TRI( v1, v2, v3 )		\
+   TriangleFunc( ctx, v1, v2, v3 )
 
-#define RENDER_QUAD( i3, i2, i1, i, pv)	\
-  draw_quad(ctx,i3,i2,i1,i,pv)
+#define RENDER_QUAD( v1, v2, v3, v4 )	\
+   QuadFunc( ctx, v1, v2, v3, v4 )
 
+#define TAG(x) _tnl_##x##_verts
 
 #define LOCAL_VARS							\
-    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb; \
-    (void)VB;
+    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;			\
+    const GLuint * const elt = VB->Elts;				\
+    const line_func LineFunc = ctx->Driver.LineFunc;			\
+    const triangle_func TriangleFunc = ctx->Driver.TriangleFunc;	\
+    const quad_func QuadFunc = ctx->Driver.QuadFunc;			\
+    (void) (LineFunc && TriangleFunc && QuadFunc);			\
+    (void) elt;
 
-#define TAG(x) x##_clipped
 #define RESET_STIPPLE ctx->Driver.ResetLineStipple( ctx )
 #define RESET_OCCLUSION ctx->OcclusionResult = GL_TRUE;
+#define RENDER_TAB_QUALIFIER 
 #define PRESERVE_VB_DEFS
 #include "t_vb_rendertmp.h"
 
 
-
-/* Elts, with the possibility of clipping.
+/* Elts, no clipping.
  */
 #undef ELT
-#undef TAG
-#undef LOCAL_VARS
+#define TAG(x) _tnl_##x##_elts
 #define ELT(x) elt[x]
-#define LOCAL_VARS						\
-    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;		\
-    const GLuint * const elt = VB->Elts;			\
-    (void) elt;
-#define TAG(x) x##_clipped_elts
-
 #include "t_vb_rendertmp.h"
 
 
 
+
 /**********************************************************************/
 /*              Clip and render whole vertex buffers                  */
 /**********************************************************************/
@@ -520,40 +568,29 @@ static GLboolean run_render( GLcontext *ctx,
    render_func *tab;
    GLint pass = 0;
 
-/*     return GL_FALSE; */
+   VB->interpfunc = RENDER_STAGE_DATA(stage)->interp;
+   VB->copypvfunc = RENDER_STAGE_DATA(stage)->copypv;
 
-   VB->interpfunc = (void *)RENDER_STAGE_DATA(stage)->interp;
+   /* Allow the drivers to lock before projected verts are built so
+    * that window coordinates are guarenteed not to change before
+    * rendering.
+    */
+   if (ctx->Driver.RenderStart)
+      ctx->Driver.RenderStart( ctx );
    
-   if (new_inputs) {
-      GLuint importable = new_inputs & VB->importable_data;
-      GLuint interested = 0;
-
-      if (VB->ClipOrMask)
-	 interested = ~0;
-      
-      if (ctx->_TriangleCaps & DD_TRI_UNFILLED) 
-	 interested |= VERT_EDGE;
-      
-      importable &= interested;
-
-      if (importable) 
-	 VB->import_data( ctx, importable, VEC_NOT_WRITEABLE|VEC_BAD_STRIDE);
-
-      if (ctx->Driver.BuildProjectedVertices)
-	 ctx->Driver.BuildProjectedVertices( ctx, 0, VB->Count, new_inputs);
-   }
+   if (VB->ClipOrMask) {
+      tab = VB->Elts ? clip_render_tab_elts : clip_render_tab_verts;
 
-   /* Rendering is considered a side-effect, and must be repeated each
-    * time the stage is run, even if no inputs have changed.
-    */
-   if (VB->Elts) { 
-      tab = VB->ClipOrMask ? render_tab_clipped_elts : render_tab_raw_elts;
-   } else {
-      tab = VB->ClipOrMask ? render_tab_clipped : render_tab_raw;
+      if (new_inputs & VB->importable_data) 
+	 VB->import_data( ctx,
+			  new_inputs & VB->importable_data,
+			  VEC_NOT_WRITEABLE|VEC_BAD_STRIDE);
    }
+   else {
+      tab = VB->Elts ? ctx->Driver.RenderTabElts : ctx->Driver.RenderTabVerts;
+   } 
 
-   if (ctx->Driver.RenderStart)
-      ctx->Driver.RenderStart( ctx );
+   ctx->Driver.BuildProjectedVertices( ctx, 0, VB->Count, new_inputs );
 
    do
    {
@@ -564,9 +601,6 @@ static GLboolean run_render( GLcontext *ctx,
 	 length= VB->PrimitiveLength[i];	
 	 ASSERT(length || (flags & PRIM_LAST));
 	 ASSERT((flags & PRIM_MODE_MASK) <= GL_POLYGON+1);
-/*  	 fprintf(stderr, "render %s %d..%d\n",  */
-/*  		 _mesa_prim_name[flags & PRIM_MODE_MASK], */
-/*  		 i, i+length); */
 	 if (length)
 	    tab[flags & PRIM_MODE_MASK]( ctx, i, i + length, flags );
       }
@@ -595,6 +629,7 @@ static void check_render( GLcontext *ctx, struct gl_pipeline_stage *stage )
 {
    struct render_stage_data *store = RENDER_STAGE_DATA(stage);
    GLuint interp = 0;
+   GLuint copy = 0;
    GLuint inputs = VERT_CLIP;
    GLuint i;
 
@@ -643,6 +678,12 @@ static void check_render( GLcontext *ctx, struct gl_pipeline_stage *stage )
       inputs |= VERT_TEX_ANY;
    }
 
+   if (ctx->_TriangleCaps & DD_FLATSHADE) {
+      copy = interp & (INTERP_RGBA|INTERP_SPEC|INTERP_INDEX);
+      interp &= ~copy;
+   }
+
+   store->copypv = copy_tab[copy];
    store->interp = interp_tab[interp];
    stage->inputs = inputs;
 }
@@ -688,6 +729,7 @@ const struct gl_pipeline_stage _tnl_render_stage =
    "render",
    (_NEW_BUFFERS |
     _DD_NEW_SEPERATE_SPECULAR |
+    _DD_NEW_FLATSHADE |
     _NEW_TEXTURE|
     _NEW_LIGHT|
     _NEW_POINT|
diff --git a/src/mesa/tnl/t_vb_rendertmp.h b/src/mesa/tnl/t_vb_rendertmp.h
index 5712d342d2..9c9271ff22 100644
--- a/src/mesa/tnl/t_vb_rendertmp.h
+++ b/src/mesa/tnl/t_vb_rendertmp.h
@@ -1,4 +1,4 @@
-/* $Id: t_vb_rendertmp.h,v 1.3 2000/12/28 22:11:06 keithw Exp $ */
+/* $Id: t_vb_rendertmp.h,v 1.4 2001/01/05 02:26:49 keithw Exp $ */
 
 /*
  * Mesa 3-D graphics library
@@ -60,6 +60,10 @@
 #define ELT(x) x
 #endif
 
+#ifndef RENDER_TAB_QUALIFIER
+#define RENDER_TAB_QUALIFIER static
+#endif
+
 static void TAG(render_points)( GLcontext *ctx,
 				GLuint start,
 				GLuint count,
@@ -68,7 +72,6 @@ static void TAG(render_points)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    RESET_OCCLUSION;
    INIT(GL_POINTS);
    RENDER_POINTS( start, count );
@@ -84,7 +87,6 @@ static void TAG(render_lines)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    RESET_OCCLUSION;
    INIT(GL_LINES);
    for (j=start+1; j<count; j+=2 ) {
@@ -104,7 +106,6 @@ static void TAG(render_line_strip)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    RESET_OCCLUSION;
    INIT(GL_LINES);
 
@@ -128,7 +129,6 @@ static void TAG(render_line_loop)( GLcontext *ctx,
 
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    RESET_OCCLUSION;
    INIT(GL_LINES);
 
@@ -160,18 +160,17 @@ static void TAG(render_triangles)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    INIT(GL_POLYGON);
    if (NEED_EDGEFLAG_SETUP) {
       for (j=start+2; j<count; j+=3) {
 	 /* Leave the edgeflags as supplied by the user.
 	  */
-	 RENDER_TRI( ELT(j-2), ELT(j-1), ELT(j), ELT(j), 0 );
+	 RENDER_TRI( ELT(j-2), ELT(j-1), ELT(j) );
 	 RESET_STIPPLE;
       }
    } else {
       for (j=start+2; j<count; j+=3) {
-	 RENDER_TRI( ELT(j-2), ELT(j-1), ELT(j), ELT(j), 0 );
+	 RENDER_TRI( ELT(j-2), ELT(j-1), ELT(j) );
       }
    }
    POSTFIX;
@@ -191,15 +190,11 @@ static void TAG(render_tri_strip)( GLcontext *ctx,
    if (TEST_PRIM_PARITY(flags))
       parity = 1;
    
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    INIT(GL_POLYGON);
    if (NEED_EDGEFLAG_SETUP) {
       for (j=start+2;j<count;j++,parity^=1) {
-	 /* All edges are boundary.  Set edgeflags to 1, draw the
-	  * triangle, and restore them to the original values.
-	  */
-	 GLuint ej2 = ELT(j-2);
-	 GLuint ej1 = ELT(j-1);
+	 GLuint ej2 = ELT(j-2+parity);
+	 GLuint ej1 = ELT(j-1-parity);
 	 GLuint ej = ELT(j);
 	 GLboolean ef2 = EDGEFLAG_GET( ej2 );
 	 GLboolean ef1 = EDGEFLAG_GET( ej1 );
@@ -207,15 +202,15 @@ static void TAG(render_tri_strip)( GLcontext *ctx,
 	 EDGEFLAG_SET( ej2, GL_TRUE );
 	 EDGEFLAG_SET( ej1, GL_TRUE );
 	 EDGEFLAG_SET( ej, GL_TRUE );
-	 RENDER_TRI( ej2, ej1, ej, ej, parity );
+	 RENDER_TRI( ej2, ej1, ej );
 	 EDGEFLAG_SET( ej2, ef2 );
 	 EDGEFLAG_SET( ej1, ef1 );
 	 EDGEFLAG_SET( ej, ef );
 	 RESET_STIPPLE;
       }
    } else {
-      for (j=start+2;j<count;j++,parity^=1) {
-	 RENDER_TRI( ELT(j-2), ELT(j-1), ELT(j), ELT(j), parity );
+      for (j=start+2; j<count ; j++, parity^=1) {
+	 RENDER_TRI( ELT(j-2+parity), ELT(j-1-parity), ELT(j) );
       }
    }
    POSTFIX;
@@ -231,7 +226,6 @@ static void TAG(render_tri_fan)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    INIT(GL_POLYGON);
    if (NEED_EDGEFLAG_SETUP) {
       for (j=start+2;j<count;j++) {
@@ -246,7 +240,7 @@ static void TAG(render_tri_fan)( GLcontext *ctx,
 	 EDGEFLAG_SET( ejs, GL_TRUE );
 	 EDGEFLAG_SET( ej1, GL_TRUE );
 	 EDGEFLAG_SET( ej, GL_TRUE );
-	 RENDER_TRI( ejs, ej1, ej, ej, 0);
+	 RENDER_TRI( ejs, ej1, ej);
 	 EDGEFLAG_SET( ejs, efs );
 	 EDGEFLAG_SET( ej1, ef1 );
 	 EDGEFLAG_SET( ej, ef );
@@ -254,7 +248,7 @@ static void TAG(render_tri_fan)( GLcontext *ctx,
       }
    } else {
       for (j=start+2;j<count;j++) {
-	 RENDER_TRI( ELT(start), ELT(j-1), ELT(j), ELT(j), 0 );
+	 RENDER_TRI( ELT(start), ELT(j-1), ELT(j) );
       }
    }
 
@@ -262,22 +256,15 @@ static void TAG(render_tri_fan)( GLcontext *ctx,
 }
 
 
-/* This is a bit of a hack.  Clipping produces polygons and really
- * wants to use this function to render them (in particular to get the
- * edgeflags right).  However, the rule that pv==start for polys
- * doens't hold there, hence the extra arg and the wrapper below.  
- */
-static void TAG(render_poly_pv)( GLcontext *ctx,
-				 GLuint start,
-				 GLuint count,
-				 GLuint flags,
-				 GLuint pv )
+static void TAG(render_poly)( GLcontext *ctx,
+			      GLuint start,
+			      GLuint count,
+			      GLuint flags )
 {
    GLuint j = start+2;
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    INIT(GL_POLYGON);
    if (NEED_EDGEFLAG_SETUP) {
       GLboolean efstart = EDGEFLAG_GET( ELT(start) );
@@ -300,7 +287,7 @@ static void TAG(render_poly_pv)( GLcontext *ctx,
       if (j<count-1) {
 	 GLboolean ef = EDGEFLAG_GET( ELT(j) );
 	 EDGEFLAG_SET( ELT(j), GL_FALSE );
-	 RENDER_TRI( ELT(start), ELT(j-1), ELT(j), ELT(pv), 0 );
+	 RENDER_TRI( ELT(start), ELT(j-1), ELT(j) );
 	 EDGEFLAG_SET( ELT(j), ef );
 	 j++;
 	    
@@ -311,7 +298,7 @@ static void TAG(render_poly_pv)( GLcontext *ctx,
 	 for (;j<count-1;j++) {
 	    GLboolean efj = EDGEFLAG_GET( ELT(j) );
 	    EDGEFLAG_SET( ELT(j), GL_FALSE );
-	    RENDER_TRI( ELT(start), ELT(j-1), ELT(j), ELT(pv), 0 );
+	    RENDER_TRI( ELT(start), ELT(j-1), ELT(j) );
 	    EDGEFLAG_SET( ELT(j), efj );
 	 }
       }
@@ -319,7 +306,7 @@ static void TAG(render_poly_pv)( GLcontext *ctx,
       /* Draw the last or only triangle
        */
       if (j < count)
-	 RENDER_TRI( ELT(start), ELT(j-1), ELT(j), ELT(pv), 0 );
+	 RENDER_TRI( ELT(j-1), ELT(j), ELT(start) );
 
       /* Restore the first and last edgeflags:
        */
@@ -332,21 +319,12 @@ static void TAG(render_poly_pv)( GLcontext *ctx,
    }
    else {
       for (j=start+2;j<count;j++) {
-	 RENDER_TRI( ELT(start), ELT(j-1), ELT(j), ELT(start), 0 );
+	 RENDER_TRI( ELT(j-1), ELT(j), ELT(start) );
       }
    }
    POSTFIX;
 }
 
-static void TAG(render_poly)( GLcontext *ctx,
-			      GLuint start,
-			      GLuint count,
-			      GLuint flags )
-{
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
-   TAG(render_poly_pv)( ctx, start, count, flags, start );
-}
-
 static void TAG(render_quads)( GLcontext *ctx,
 			       GLuint start,
 			       GLuint count,
@@ -356,18 +334,17 @@ static void TAG(render_quads)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    INIT(GL_POLYGON);
    if (NEED_EDGEFLAG_SETUP) {
       for (j=start+3; j<count; j+=4) {
 	 /* Use user-specified edgeflags for quads.
 	  */
-	 RENDER_QUAD( ELT(j-3), ELT(j-2), ELT(j-1), ELT(j), ELT(j) );
+	 RENDER_QUAD( ELT(j-3), ELT(j-2), ELT(j-1), ELT(j) );
 	 RESET_STIPPLE;
       }
    } else {
       for (j=start+3; j<count; j+=4) {
-	 RENDER_QUAD( ELT(j-3), ELT(j-2), ELT(j-1), ELT(j), ELT(j) );
+	 RENDER_QUAD( ELT(j-3), ELT(j-2), ELT(j-1), ELT(j) );
       }
    }
    POSTFIX;
@@ -382,7 +359,6 @@ static void TAG(render_quad_strip)( GLcontext *ctx,
    LOCAL_VARS;
    (void) flags;
 
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    INIT(GL_POLYGON);
    if (NEED_EDGEFLAG_SETUP) {
       for (j=start+3;j<count;j+=2) {
@@ -397,7 +373,7 @@ static void TAG(render_quad_strip)( GLcontext *ctx,
 	 EDGEFLAG_SET( ELT(j-2), GL_TRUE );
 	 EDGEFLAG_SET( ELT(j-1), GL_TRUE );
 	 EDGEFLAG_SET( ELT(j), GL_TRUE );
-	 RENDER_QUAD( ELT(j-3), ELT(j-2), ELT(j), ELT(j-1), ELT(j) );
+	 RENDER_QUAD( ELT(j-1), ELT(j-3), ELT(j-2), ELT(j) );
 	 EDGEFLAG_SET( ELT(j-3), ef3 );
 	 EDGEFLAG_SET( ELT(j-2), ef2 );
 	 EDGEFLAG_SET( ELT(j-1), ef1 );
@@ -406,7 +382,7 @@ static void TAG(render_quad_strip)( GLcontext *ctx,
       }
    } else {
       for (j=start+3;j<count;j+=2) {
-	 RENDER_QUAD( ELT(j-3), ELT(j-2), ELT(j), ELT(j-1), ELT(j) );
+	 RENDER_QUAD( ELT(j-1), ELT(j-3), ELT(j-2), ELT(j) );
       }
    }
    POSTFIX;
@@ -417,14 +393,13 @@ static void TAG(render_noop)( GLcontext *ctx,
 			      GLuint count,
 			      GLuint flags )
 {
-/*     fprintf(stderr, "%s %d..%d\n", __FUNCTION__, start, count); */
    (void)(ctx && start && count && flags);
 }
 
-static void (*TAG(render_tab)[GL_POLYGON+2])(GLcontext *,
-					     GLuint,
-					     GLuint,
-					     GLuint) = 
+RENDER_TAB_QUALIFIER void (*TAG(render_tab)[GL_POLYGON+2])(GLcontext *,
+							   GLuint,
+							   GLuint,
+							   GLuint) = 
 {
    TAG(render_points),
    TAG(render_lines),
@@ -452,6 +427,7 @@ static void (*TAG(render_tab)[GL_POLYGON+2])(GLcontext *,
 #undef RESET_STIPPLE
 #undef DBG
 #undef ELT
+#undef RENDER_TAB_QUALIFIER
 #endif
 
 #ifndef PRESERVE_TAG