>> A nice step towards fixing pr16187 :-) > > I'm not sure I fully understand the details of that bug. Are you > thinking we could solve it if we delay more kinds of declarations so > the visibility gets computed at the end? >
Correct. To fix it we would have to delay the computation of any visibility based on a decl that can go from implicitly hidden (-fvisibility=hidden) to explicitly default (attribute). With that we could then cache the visibility computation, which would hopefully pay for the cost of delaying the computation. Cheers, Rafael _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
