This includes
- doing prescan of shader to know version/profile before parsing it
- putting precision qualifiers on built-in ES symbols
- getting most built-in state correct for core/compatibility/missing profile
- adding gl_VertexID and gl_InstanceID, among other ES 300 built-in symbols
- adding the ES 300 gl_Max/Min constants
- accepting shaders that contain nothing but whitespace without generating an error
git-svn-id: https://cvs.khronos.org/svn/repos/ogl/trunk/ecosystem/public/sdk/tools/glslang@20627 e7fa87d3-cd2b-0410-9028-fcbf551c1848
For 4.2, largely, it is only the grammar that is working. Productions and semantics are mostly missing. Lexical analysis is mostly done, but not in the preprocessor, which still can't handle uint and double literals.
The grammar and token names are reorganized to match the specification, to allow easier comparison between the specification and the working grammar.
git-svn-id: https://cvs.khronos.org/svn/repos/ogl/trunk/ecosystem/public/sdk/tools/glslang@19946 e7fa87d3-cd2b-0410-9028-fcbf551c1848