Chris Forbes <[email protected]> writes: > +static bool > +test_vertex_attribs() > +{ > + bool pass = true; > + GLuint data = 0; > + > + /* "The error INVALID_OPERATION is generated by ... , or > VertexAttribPointer > + * if <type> is UNSIGNED_INT_10F_11F_11F_REV and <size> is not 3. > + */ > + > + glVertexAttribPointer(0, 3, GL_UNSIGNED_INT_10F_11F_11F_REV, GL_FALSE, > 0, (GLvoid *)0); > + TEST("VertexAttribPointer-ok", GL_NO_ERROR); > + > + glVertexAttribPointer(0, 2, GL_UNSIGNED_INT_10F_11F_11F_REV, GL_FALSE, > 0, (GLvoid *)0); > + TEST("VertexAttribPointer-badsize", GL_INVALID_OPERATION);
I think file could also use a test that a normalized == GL_TRUE doesn't
spuriously throw an error, given how much spec text is devoted to that.
> +void
> +piglit_init(int argc, char **argv)
> +{
> + bool pass = true;
> + GLuint bo;
> +
> + piglit_require_extension("GL_ARB_vertex_type_2_10_10_10_rev");
> + piglit_require_extension("GL_ARB_vertex_type_10f_11f_11f_rev");
The 2_10_10_10 check seems gratuitous. In particular, if an
implementation exposed 10f without the infrastructure from 2_10, you'd
want this test to run and fail.
Other than that,
Reviewed-by: Eric Anholt <[email protected]>
pgppsPSVOgvWt.pgp
Description: PGP signature
_______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
