<p dir="ltr"><br>
On Nov 27, 2015 10:21 AM, "Emil Velikov" <<a href="mailto:emil.l.velikov@gmail.com">emil.l.velikov@gmail.com</a>> wrote:<br>
><br>
> On 27 November 2015 at 18:00, Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> wrote:<br>
> ><br>
> > On Nov 25, 2015 1:27 PM, "Emil Velikov" <<a href="mailto:emil.l.velikov@gmail.com">emil.l.velikov@gmail.com</a>> wrote:<br>
> >><br>
> >> From: Emil Velikov <<a href="mailto:emil.velikov@collabora.com">emil.velikov@collabora.com</a>><br>
> >><br>
> >> ... alongside its C++ brethren. On the flip side the files are called<br>
> >> nir_types.{cpp,h} just because...<br>
> ><br>
> > It does do one NIR-specific thing: if you call glsl_get_length on a matrix,<br>
> > it gives you the number of columns. This probably isn't a huge deal though.<br>
> ><br>
> By "NIR-specific" I believe you mean "used solely by NIR" or am I<br>
> missing something ?</p>
<p dir="ltr">I mean that GLSL IR and NIR treat matrices differently. NIR has no concept of a matrix but rather treats them as an array of vectors. GLSL IR treats matrices and vectors more-or-less the same; matrices just have more than one column. However, almost all C code that works with glsl_type is working with NIR so this shouldn't be a problem.<br>
</p>