On 19 September 2011 09:50, Eric Anholt <span dir="ltr"><<a href="mailto:eric@anholt.net">eric@anholt.net</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On Thu, 15 Sep 2011 09:16:39 -0700, Paul Berry <<a href="mailto:stereotype441@gmail.com">stereotype441@gmail.com</a>> wrote:<br>
> This patch adds two tests that exercise a subtle corner case bug found<br>
> in Mesa with i965: If a value is negated, and then negated again, then<br>
> copy propagation would fail to un-do the negation.<br>
><br>
> The bug only exists on i965 vertex shaders, but I went ahead and<br>
> added the equivalent fragment shader test for completeness.<br>
<br>
</div>For bugs where I was definitely trying to trigger a copy-propagation<br>
bug, I was naming them glsl-*copy-propagation before. Not a big deal<br>
either way. Thanks for the tests!<br>
</blockquote></div><br>Sounds reasonable to me. I'll rename the tests and move them to the "shaders" directory with the other copy propagation tests when I push the patches.<br>