On 15 January 2014 02:46, Tapani <tapani.pa...@intel.com> wrote: > On 01/14/2014 07:53 PM, Paul Berry wrote: > > On 2 January 2014 03:58, Tapani Pälli <tapani.pa...@intel.com> wrote: > >> +static void >>> +serialize_uniform_storage(gl_uniform_storage *uni, memory_writer &blob) >>> >> >> I don't think this is right. The ARB_get_program_binary spec says: >> >> 8. How does ProgramBinary interact with uniform values, including >> shader-specified defaults? >> >> RESOLVED: All uniforms specified in the binary are reset to their >> shader- >> specified defaults, or to zero if they aren't specified, when the >> program >> binary is loaded. The spec language has been updated to specify this >> behavior. >> >> The OES_get_program_binary spec doesn't mention uniforms at all, but I >> believe this is not intended to indicate a behavioural difference--it's >> just a consequence of the fact that ARB_get_program_binary was written >> later, so they had a chance to clarify things. In fact, issue #7 in >> ARB_get_program_binary specifically says: >> >> 7. How does this spec differ from the OES_get_program_binary and why? >> >> ... >> >> There are other areas where language was clarified, but this is meant >> to >> be consistent with the intent of the original specification and not to >> alter previously established behavior. >> >> So I believe we shouldn't be serializing uniform values. >> >> >> For me this seemed much easier way to serialize than recreate it though. > Would it be enough if I reset the default values in place? >
Yes, I think that would be reasonable. Thanks!
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
