Author: rambetter

New Revision: 376
Undoing revision 375 because it did not offer measured accuracy improvements
in the q3map2 math calculations.

diff --git a/regression_tests/q3map2/disappearing_sliver2/README.txt b/regression_tests/q3map2/disappearing_sliver2/README.txt
index c0d4eaa687df1e203dfa9b2b9e60cfa0019a63a6..55f9b51dbe3b8b4a94f77bdf1df36205831c619d 100644 (file)
--- a/regression_tests/q3map2/disappearing_sliver2/README.txt
+++ b/regression_tests/q3map2/disappearing_sliver2/README.txt
@@ -66,3 +66,18 @@ is this:
 
 As you can see, all points but one have an increase in accuracy.  This is
 still not accurate enough in my opinion, but is a step in the right direction.
+
+
+MORE NOTES:
+===========
+
+I attempted to improve upon revision 371 by streamlining the code in
+BaseWindingForPlane() some more.  Those attempts were committed as r375.
+After revision 375:
+
+  (6784.09375000 16241.01757812 -1722.04687500)
+  (6144.00000000 16082.99414062 -1443.00390625)
+  (6144.00000000 16122.00000000 -1424.00097656)
+
+Revision 375 has since been reverted (undone) because of the loss in
+accuracy.

diff --git a/tools/quake3/common/polylib.c b/tools/quake3/common/polylib.c
index 7d5b8a83c8056a900b3280d28b0a22711af956b0..b62d12ff34f762071ec6d5d17241546a54c034a3 100644 (file)
--- a/tools/quake3/common/polylib.c
+++ b/tools/quake3/common/polylib.c
@@ -253,9 +253,13 @@ winding_t *BaseWindingForPlane (vec3_t normal, vec_t dist)
                        vright[2] = normal[1];
                        break;
        }
-       // NOTE: vright is NOT a unit vector at this point.
-       VectorSetLength(vright, MAX_WORLD_COORD * 2, vright);
        CrossProduct(normal, vright, vup);
+
+       // IMPORTANT NOTE: vright and vup are NOT unit vectors at this point.
+       // However, normal, vup, and vright are pairwise perpendicular.
+
+       VectorSetLength(vup, MAX_WORLD_COORD * 2, vup);
+       VectorSetLength(vright, MAX_WORLD_COORD * 2, vright);
        VectorScale(normal, dist, org);
 
        w = AllocWinding(4);