On Oct 27, 2008, at 7:32 AM, Doug Gregor wrote: > On Mon, Oct 27, 2008 at 10:22 AM, Ted Kremenek <[EMAIL PROTECTED]> > wrote: >> >> On Oct 27, 2008, at 5:50 AM, Douglas Gregor wrote: >> >> Author: dgregor >> Date: Mon Oct 27 07:50:38 2008 >> New Revision: 58244 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=58244&view=rev >> Log: >> When destroying a translation unit, deallocate its owned >> declarations in >> reverse order, because there may be dependencies among the >> declarations. >> >> Hi Doug, >> Do you have a concrete case of why this is necessary? It seems to >> me that a >> solid ownership policy in the ASTs wouldn't require this to be in >> place. I >> feel like I'm being naive about something very basic. > > Yeah, test/SemaCXX/fntype-decl.cpp causes the problem, but you'll need > to run under valgrind to see it. Argiris describes the issue here: > > > http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20081020/008535.html > > - Doug
I see. The rabbit hole ends where TypedefType depends on TypedefDecl to get the underlying type. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
