On Mar 12, 2012, at 3:41 AM, James Molloy wrote: > Hi John, > > Attached is a slightly modified version of the patch. > >> I would prefer to only compute this when there's actually a default > argument >> on the old declaration. It should be easy enough to just pass down the > scope >> of the new declaration. > > I have done this. > >> This doesn't seem to be right. We should be ignoring the default argument >> in other scopes regardless of whether the new declaration has a default >> argument. > > While I originally agreed with you on this, I was at home, didn't have the > patch available and couldn't see that actually the patch *does* actually do > the right thing. I have added an extra testcase however as this behaviour is > not currently tested. See i() in SemaCXX/default1.cpp. > >> Also, don't we need to look for previous declarations that *are* in the > same >> scope and make sure we use their default arguments? > > Again, this wasn't tested but does do the right thing. I've added a testcase > into CodeGenCXX/default-arguments.cpp based on the snippet you posted. > > Is this OK now?
Yeah, looks good. Thanks! John. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
