On Fri, 2015-11-06 at 13:16 +0000, Emil Velikov wrote: > On 5 November 2015 at 11:17, Timothy Arceri <t_arc...@yahoo.com.au> wrote: > > From: Timothy Arceri <timothy.arc...@collabora.com> > > > > This is in preperation for compile-time constant support. > typo "preparation" > > > > > Also fix up the locations for some of the extension checking > > error messages in the parser. We now correctly give the location > > of the layout qualifier identifier rather than the integer constant. > > > > The validation is moved to two locations, for validation on variables the > > checks are moved to the ast to hir pass and for qualifiers that apply to > > the > > shader the validation is moved into glsl_parser_extras.cpp. > > > > In order to do validation at the later stage in glsl_parser_extras.cpp we > > need to temporarily add a field in ast_type_qualifier to keep track of the > > parser location, this will be removed in a following patch when we > > introduce a new type for storing the comiple-time qualifiers. > > > > Also as the set_shader_inout_layout() function in glsl parser extras is > > normally called after all validation is done we need to move the code that > > sets CompileStatus and InfoLog otherwise the newly moved error messages > > will > > be ignored. > Personally I would split the validate_layout_qualifiers() introduction > and the CompileStatus/InfoLog movement into separate patches.
The reason for not doing this in a new patch is that this is existing functionality not new functionality, doing so would regress a bunch of piglit tests. I can do it if it makes things easier to review but it should all be pushed as one. > > Feel free to ignore, if others don't share my sentiment > -Emil > _______________________________________________ > mesa-dev mailing list > mesa-dev@lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/mesa-dev _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
