On Sep 8, 2010, at 10:59 PM, John McCall wrote: >> >> I agree on this point. I can tighten up the type checking. >> In any case, point of this patch was to fix the crash in a limited case and >> not treat 'struct objc_object' as a replacement for 'id' everywhere. >> So, these are alternatives. >> >> 1. Do nothing and let the test case crash ( I think it was a manufactured >> test ). >> 2. Accept the alternative 'struct ...' as builtin type everywhere. >> 3. Issue error if it is used anywhere. >> 4.??? > > Personally I like #3 because ObjC code using 'struct objc_object' is very > likely to either not compile or have really unfortunate behavioral changes — > I think Argyrios recently hunted down something in the latter category — but > that's the sort of problem we can only suss out with SWBs. For the immediate > problem, changing FunctionArgTypesAreEqual should be enough.
FWIW, I came upon the crash while testing out how 'struct objc_object' is handled in clang. So it wasn't in 'real' code but letting in known crashes is really bad IMO. -Argiris > > John. > _______________________________________________ > cfe-commits mailing list > [email protected] > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
