[Mesa3d-dev] [ mesa3d-Bugs-441859 ] glColorMaterial fails with strips

[<no...@so...>]

Bugs item #441859, was opened at 2001-07-16 17:58
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=100003&aid=441859&group_id=3

Category: mesa-core
Group: Rendering Error
Status: Open
Resolution: None
Priority: 5
Submitted By: Michael Saunders (rms7326)
Assigned to: Nobody/Anonymous (nobody)
Summary: glColorMaterial fails with strips

Initial Comment:
Using glColorMaterial(GL_FRONT_AND_BACK,
GL_AMBIENT_AND_DIFFUSE) in conjunction with
GL_QUAD_STRIP blocks does not always color/light the
surface correctly. The first few strips shade well and
then suddenly the strip turns black. This occurs only
when lighting is used. If we turn lighting off the
strips are colored correctly. We noticed that if we
deliver a color at each vertex then glColorMaterial
seems to track lighting correctly.

I should note that this problem occurs on Intel Linux
2.2 machine but not on our Solaris 7 machine. One
difference between the two builds is that I compiled
the Solaris 7 platform so that it did not use the Sparc
ASM code (i.e. USE_SPARC_ASM is commented out in conf.h
because it won't compile otherwise). So I suppose it is
possible that the problem is in that or other platform
specific code (including the graphics card I suppose).
Our linux machine has a GForce 2 Ulta.

----------------------------------------------------------------------

>Comment By: Michael Saunders (rms7326)
Date: 2001-07-17 10:39

Message:
Logged In: YES 
user_id=217333

After further investigation I discovered that I mistated
when I said that Solaris 7 did not exhibit the problem. This
is not true and I was able to get Solaris 5 and 7 to fail as
well. I tried to reduce the problem to as small as possible.
While I was doing this I notied some interesting points.

1) If we turned off drawing of the lines strips that are
also in this plot the shading problems went away. Next I ran
a larger problem and discoverd that with larger problems it
didn't matter if I turned the line strips on or not.

2) When I linked the debug build of our application to the
release build of the Mesa 3.5 library the problem did not
exhibit itself. However when I linked the release build of
our application to the release build of the Mesa 3.5 library
the problem did exhibit itself.
Initially I had assumed that we must be sending different
instructions to Mesa with our debug and release builds. So I
turned on our logging feature which outputs the OpenGL
statements delivered to Mesa and compared the two. The diffs
were identical. This leads me to believe that the problem is
an uninitialized variable inside the code that processes
strips and tracks the color and lighting (glColorMaterial).

I have included the log output as an attachment. The
instructions are all executable so if you need to paste it
into a c program it should work fine. The log contains all
the instructions send to Mesa with their respective
arguments.

----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=100003&aid=441859&group_id=3