[cfe-dev] [PATCH] Clang Static Analyzer support for temporary destructors

[Ted Kremenek]

On May 1, 2014, at 10:16 AM, Jordan Rose <jordan_rose at apple.com> wrote:

> It sounds about right to me, but (4) is interesting. Ted and Anna and I have talked about similar things before. It's a good solution for the analyzer, but it might make the rest of the compiler unhappy. IIRC, all of the existing compiler flow-sensitive analyses do include temporary destructors right now. They probably aren't doing anything so special with them, but changing the CFG so that CXXBindTemporaryExpr can be a terminator is something that potentially affects all clients.
> 
> If we do this under a flag, what behavior should we use elsewhere? The existing thing, of visiting all the conditions again?
> 

The existing thing is fine.  It is what we understand now.

IMO, hopefully a flag would be temporary.  Ideally we’d update the compiler as well in time; having so many “views” of the CFG makes it very complex to reason about.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20140501/da35b255/attachment.html>