[cfe-commits] r161187 - in /cfe/trunk: lib/Sema/SemaPseudoObject.cpp test/SemaObjC/arc-dict-bridged-cast.m
fjahanian at apple.com
Thu Aug 2 11:41:25 PDT 2012
On Aug 2, 2012, at 11:21 AM, Jordan Rose wrote:
> That is a really ugly fixit: (__bridge __strong id <NSCopying>). At the very least we should drop the __strong, but I could see an argument for just offering (__bridge id). That's all the type info we get with CFBridgingRelease, and it's not like we need more info to type-check here (since they're already casting).
Not much I can do about ugliness, it is coming of out the API being used which is common to all other situations.
> Stepping back, this is another case where we know the ownership semantics and should only offer one of the fixits. I assume this commit will already do the right thing if CFStringCreateMutable were annotated with cf_audited_transfer / arc_cf_code_audited, but maybe we should explicitly add that to the test?
Yes, adding arc_cf_code_audited makes fixit precise. Test revised in r161194.
More information about the cfe-commits