<div dir="ltr">Appreciate it very much, but I've given up up on trying to stop folks from making instcombine into the worlds hugest ball of mud. So i'm going to resign from this.</div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Aug 29, 2017 at 6:37 AM, Sanjay Patel via Phabricator <span dir="ltr"><<a href="mailto:reviews@reviews.llvm.org" target="_blank">reviews@reviews.llvm.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">spatel added a comment.<br>
<span class=""><br>
In <a href="https://reviews.llvm.org/D37195#855385" rel="noreferrer" target="_blank">https://reviews.llvm.org/<wbr>D37195#855385</a>, @aaboud wrote:<br>
<br>
> I do not see why it is much different that passing the demandedBits map and update it recursively!<br>
<br>
<br>
</span>That may be true, but there's also an argument that doing demandedBits analysis within instcombine already went over the line. :)<br>
There have been a few recent llvm-dev threads about the responsibilities and limits of various passes. Here's one:<br>
<a href="http://lists.llvm.org/pipermail/llvm-dev/2017-May/113212.html" rel="noreferrer" target="_blank">http://lists.llvm.org/<wbr>pipermail/llvm-dev/2017-May/<wbr>113212.html</a><br>
<br>
I don't have enough experience with this, so I'll defer to others for their guidance. Just thought it was worth mentioning.<br>
<br>
<br>
<a href="https://reviews.llvm.org/D37195" rel="noreferrer" target="_blank">https://reviews.llvm.org/<wbr>D37195</a><br>
<br>
<br>
<br>
</blockquote></div><br></div>