[Piglit] [PATCH v2 4/4] Add API tests for KHR_texture_compression_astc_ldr

[Chad Versace]

On Mon 14 Sep 2015, Nanley Chery wrote:
> On Tue, Sep 8, 2015 at 11:45 AM, Chad Versace <chad.versace at intel.com>
> wrote:
> 
> > On Fri 28 Aug 2015, Nanley Chery wrote:
> > > From: Nanley Chery <nanley.g.chery at intel.com>
> > >
> > > These tests check that the GL error returned by valid and invalid API
> > calls
> > > are as defined by the spec.
> > >
> > > Signed-off-by: Nanley Chery <nanley.g.chery at intel.com>
> > > ---
> > >  .../khr_texture_compression_astc/CMakeLists.gl.txt |   1 +
> > >  .../CMakeLists.gles2.txt                           |   1 +
> > >  .../khr_compressed_astc-basic.c                    | 366
> > +++++++++++++++++++++
> > >  3 files changed, 368 insertions(+)
> > >  create mode 100644
> > tests/spec/khr_texture_compression_astc/khr_compressed_astc-basic.c


> > > +static const GLenum good_targets[] = {
> >
> > There are additional good targets: 3D and and CUBE_MAP. Please rename
> > this variable to indicate that these are "good targets" for
> > glCompressedTex*3D().
> >
> I'm not sure what you mean by additional good targets. I've renamed the
> variable to good_compressed_tex_3d_targets.
> 
> > > +   GL_TEXTURE_2D_ARRAY,
> > > +   GL_TEXTURE_CUBE_MAP_ARRAY_EXT,
> > > +   GL_TEXTURE_3D,
> > > +};

I typo'd. I meant *2D* (not 3D) and CUBE_MAP are additional good targets
for ASTC. Those targets are good for glCompressedTex*2D(). In other
words, I meanth that this array isn't a list of *all* good ASTC targets,
because it excludes 2D and CUBE_MAP.

> > > +     /* Test CompressedTexImage3D */
> > > +     glCompressedTexImage3D(GL_TEXTURE_CUBE_MAP_ARRAY_EXT, 0,
> > > +             formats[fi].fmt, 4, 3, 6, 0, 6*formats[fi].bb,
> > fake_tex_data);
> > > +     REQUIRE_ERROR(GL_INVALID_VALUE);
> >
> > If HDR is unsupported, I think glCompressedTexImage3D is allowed to
> > emit GL_INVALID_OPERATION here. So, I think glCompressedTexImage3D()
> > should be tested on non-square dimensions only if HDR is supported.
> >
> Agreed. I just made a patch for mesa to support this behavior. I'm
> thinking it'd be better to check that for GL_INVALID_OPERATION if HDR is
> not supported, and GL_INVALID_VALUE otherwise.

Sounds good to me.

...

Also, the reply-level was corrupted in your message. Maybe your email
program is configured incorrectly? Perhaps its a word-wrap problem? The
bug was that the first line of each of your responses was at
reply-level=1 instead of reply-level=0.  That rendered your single-line
responses, nearly invisible, since the single line looked like it
belonged to my message instead of yours.