<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p><a class="moz-txt-link-freetext" href="https://gitlab.freedesktop.org/mesa/mesa/merge_requests/237">https://gitlab.freedesktop.org/mesa/mesa/merge_requests/237</a></p>
<div class="wiki">
<p dir="auto">New version of the thread that I sent recently,
showing two initial versions to solve the regression I found on
MR #144:</p>
<p dir="auto"><a
href="https://lists.freedesktop.org/archives/mesa-dev/2019-February/214808.html"
rel="nofollow noreferrer noopener" target="_blank">https://lists.freedesktop.org/archives/mesa-dev/2019-February/214808.html</a></p>
<p dir="auto">This MR includes a v2 of the second option. It
includes Jason's feedback plus some extra cleaning-ups that I
found while re-checking the GLSL linker (like removing
PixelCenterInteger/OriginUpperLeft from gl_program, as that info
is tracked at gl_program.info.fs.xxx). Due that, it touches a
lot of places on Mesa.</p>
<p dir="auto">This MR also includes a second patch, that removes
pixel_center_integer/origin_upper_left from the ir variable.
Basically because that is already tracked in a lot of places, so
it is not really needed. It is on a different patch because I
initially though of it as something optional, as it is not
really required to fix the regression. But after finishing it,
the only reason to keep it as a different patch is to make
easier the review. I think that it would be better to squash
both patches, although I don't have a strong opinion, I would
let the reviewer give his opinion.</p>
</div>
<p><br>
</p>
</body>
</html>