<div dir="auto"><div><div class="gmail_extra"><div class="gmail_quote">On 21 Mar 2017 12:23 pm, "Chandler Carruth" <<a href="mailto:chandlerc@gmail.com">chandlerc@gmail.com</a>> wrote:<br type="attribution"><blockquote class="quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div class="gmail_quote"><div class="quoted-text"><div dir="ltr">On Tue, Mar 21, 2017 at 6:06 PM Hal Finkel via cfe-dev <<a href="mailto:cfe-dev@lists.llvm.org" target="_blank">cfe-dev@lists.llvm.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div bgcolor="#FFFFFF" text="#000000" class="m_-8074631411299533964gmail_msg">
<p class="m_-8074631411299533964gmail_msg"><br class="m_-8074631411299533964gmail_msg">
</p>
<div class="m_-8074631411299533964m_-4675691131510924527moz-cite-prefix m_-8074631411299533964gmail_msg">On 03/21/2017 08:45 AM, Hubert Tong
wrote:<br class="m_-8074631411299533964gmail_msg">
</div>
<blockquote type="cite" class="m_-8074631411299533964gmail_msg">
<div dir="ltr" class="m_-8074631411299533964gmail_msg">
<div class="m_-8074631411299533964gmail_msg">This question was explored in an other context here:<br class="m_-8074631411299533964gmail_msg">
<a href="http://sourcerytools.com/pipermail/cxx-abi-dev/2013-January/002544.html" class="m_-8074631411299533964gmail_msg" target="_blank">http://sourcerytools.com/<wbr>pipermail/cxx-abi-dev/2013-<wbr>January/002544.html</a><br class="m_-8074631411299533964gmail_msg">
</div>
</div>
</blockquote>
<br class="m_-8074631411299533964gmail_msg"></div><div bgcolor="#FFFFFF" text="#000000" class="m_-8074631411299533964gmail_msg">
Very interesting, thanks! So the conclusion was that, "
The ABI will need to specify the layout of closure types with weak
linkage."</div></blockquote><div><br></div></div><div>Are you basing that on what the document does or the discussion in the thread? Because I don't really see a conclusion in the thread. I see John pointing out three options and suggestion one which is not what you claim is the conclusion here:</div><div><a href="http://sourcerytools.com/pipermail/cxx-abi-dev/2013-January/002544.html" target="_blank">http://sourcerytools.com/<wbr>pipermail/cxx-abi-dev/2013-<wbr>January/002544.html</a><br></div><div><br></div><div>I don't see any real disagreement but I also don't see any real discussion past that email and brief follow-ups....</div></div></div>
</blockquote></div><br></div></div><div class="gmail_extra" dir="auto">I expect John's #1 is most likely to be the eventual outcome; #2 and #3 are too restrictive. Plus the language is moving in a direction that makes the capture set much easier to reliably compute, so the cost of #1 is going to become much more like "don't do those optimisations you weren't really doing much anyway" and much less "accurately implement these arcane and formally unobservable rules or you violate the ABI".</div></div>