<div dir="ltr">So is this OK to go in?<div><br></div><div>Stuart</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Feb 1, 2013 at 3:26 PM, Stuart Abercrombie <span dir="ltr"><<a href="mailto:sabercrombie@google.com" target="_blank">sabercrombie@google.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Re-adding the list I removed by mistake.<div><br></div><div>I agree we might end up with cases where we want to have a different #version, but my assumption is that this would be the exception rather than the norm. We can add an override for inserting some other version later if this becomes useful.</div>
<div><br></div><div>I just want to avoid explicit #version directives unnecessarily tying tests to a particular API, as they do right now. I can't see a good way to stop this happening except by making #version directives only appear in the shader source as the result of a more intentional decision than directly including them in the shaders.</div>
<span class="HOEnZb"><font color="#888888">
<div><br></div><div>Stuart</div></font></span><div><div class="h5"><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Feb 1, 2013 at 10:29 AM, Tom Gall <span dir="ltr"><<a href="mailto:tom.gall@linaro.org" target="_blank">tom.gall@linaro.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Stuart,<br>
<br>
I like your plan.<br>
<div><br>
On Fri, Feb 1, 2013 at 10:48 AM, Stuart Abercrombie<br>
<<a href="mailto:sabercrombie@google.com" target="_blank">sabercrombie@google.com</a>> wrote:<br>
> The aim in my mind is definitely commonality. I want to exploit that<br>
> reusabilty.<br>
><br>
> The change I made recently to the GLSL ES requirement syntax, and really the<br>
> goal of a lot of what I've been doing, was to allow specification of<br>
> different GLSL versions for regular and ES GL in the same file. That's not<br>
> there yet, but I think it should be fairly simple to do.<br>
><br>
> My plan is:<br>
><br>
> 1. Ensure we have GLSL requirements everywhere (for now, either GLSL or GLSL<br>
> ES, but not both).<br>
<br>
</div>Least for the GLSL ES case it's pretty nonexistent least for the one<br>
testcase in tree and the boatload I'm sitting on.<br>
<div><br>
> 2. Remove explicit #version directives from all shader_test files and ban<br>
> them in shader_runner. They will then always be inserted by shader_runner<br>
> based on the GLSL requirement.<br>
<br>
</div>I'm not sure that #version should always be inserted. After all I can<br>
see situations where for a glsl 1.10 testcase, you would want to run<br>
that same code on a newer version of glsl.<br>
<div><br>
> 3. Change shader_runner to understand simultaneous GLSL and GLSL ES<br>
> requirements, and insert the relevant #version directive depending on the<br>
> API in use.<br>
<br>
</div>Yup!<br>
<div><br>
> Does this seem reasonable? I think we're pursuing the same goal.<br>
<br>
</div>It sure does and I'm very glad to hear it.<br>
<br>
> Stuart<br>
<br>
<snip><br>
<div><div><br>
--<br>
Regards,<br>
Tom<br>
<br>
"Where's the kaboom!? There was supposed to be an earth-shattering<br>
kaboom!" Marvin Martian<br>
Tech Lead, Graphics Working Group | Linaro.org │ Open source software<br>
for ARM SoCs<br>
w) tom.gall att <a href="http://linaro.org" target="_blank">linaro.org</a><br>
h) tom_gall att <a href="http://mac.com" target="_blank">mac.com</a><br>
</div></div></blockquote></div><br></div></div></div></div>
</blockquote></div><br></div>