On Thu, Jul 30, 2009 at 10:35 AM, Ian Romanick<[email protected]> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Brian Paul wrote: >> On Wed, Jul 29, 2009 at 7:34 PM, Ian Romanick<[email protected]> wrote: >>> >>> I've noticed an odd behavior in Mesa's current shader assembler. Given >>> a program like progs/vp/arl-static.txt, it will emit an error message >>> but assemble and use the program. Is this intentional? Are there some >>> real-world applications that use incorrect shaders that just happen to >>> work on some implementations? >>> >>> The reason that I'm asking is my new assembler does not reproduce this >>> behavior. It adheres to the spec and rejects this program. >>> >>> In the code from arl-static.txt reproduced below, "arr[5]" should be >>> "arr[7]". >>> >>> !!ARBvp1.0 >>> PARAM arr[5] = { {.0,.1,.2,.3}, {.4,.5,.6,.7}, {.8,.9,.10,.1}, >>> {.12,.3,.4,.14}, {.5,.8,.1,.9}, {.2,0,.4,.1}, {.6,.1,.8,.9}}; >>> ADDRESS addr; >>> ARL addr.x, {3}.x; >>> MOV result.color, arr[addr.x]; >>> MOV result.position, vertex.position; >>> END >> >> The attached patch should fix this. Please commit if you agree. > > I'll do some testing on it, but it looks like it should give spec > adherent behavior. I mainly wanted to be sure this wasn't intentional. > There's at least one other place (the WoW vertex.weight hack) where we > intentionally deviate from the spec to work around an app bug. I wanted > to be sure this wasn't another case of that.
It was a simple bug in this case. I've committed the fix. -Brian ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Mesa3d-dev mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
