<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class=""><br class=""><div class="">Sanjoy,</div><div class=""> in your blog post <a href="https://www.playingwithpointers.com/problem-with-undef.html" class="">https://www.playingwithpointers.com/problem-with-undef.html</a></div><div class="">you describe a problem with LLVM “undef”,</div><div class="">yet in your paper <a href="http://www.cs.utah.edu/~regehr/papers/undef-pldi17.pdf" class="">http://www.cs.utah.edu/~regehr/papers/undef-pldi17.pdf</a></div><div class="">you do not suggest fixing this problem, even though in chpt. 9 you identify other</div><div class="">compilers that do attempt to avoid it.</div><div class=""><br class=""></div><div class="">It seems to me that since transforming a program with multiple uses of an “undef” variable</div><div class="">into one with multiple distinct “undef”s results in behavior that is inconsistent (and that</div><div class="">violates users expectations) we should not allow it.</div><div class=""><br class=""></div><div class="">This seems to me to be one of those situations where the phrase “extraordinary claims</div><div class="">require extraordinary evidence” applies, and I’m not seeing any, are you ?</div><div class=""><br class=""></div><div class=""><br class=""></div><div class=""><div class="">IIUC, a “freeze” operation can always be inserted over an “undef” (now “poison”)</div><div class="">that forces a single consistent value, but why the extra overhead of “freeze”, as</div><div class="">there doesn’t seem to be a convincing argument for ever allowing the same variable</div><div class="">to have inconsistent values, or am I missing something ?</div></div><div class=""><br class=""></div><div class=""><br class=""></div><div class="">Peter Lawrence.</div></body></html>