> Am 23.02.2016 um 10:14 schrieb David Chisnall <[email protected]>: > > Have you also looked at adding instancetype to the various places that expect > it and at the nullability attributes? The latter is primarily intended for > Swift interop, but does make the static analyser results more useful (and can > generate better code in a few places, though not by much).
I have added backwards-compatibility definitions of instancetype (defined to id if unsupported by the compiler) and updated the collection classes I was going over anyways. This promptly caught two (albeit harmless) cases in base where an immutable object was assigned to a variable typed for the mutable subclass. Regarding nullability annotations, I've already added the macros for non-null audited sections, but I'm still a bit unsure about how to handle/define the nullability qualifiers, especially since there are now three variants of them (the double-underscore ones, the underscore-and-capitalised ones, and the contextual variants without underscores in method parameters). So yeah, it’s on my list. Cheers, Niels _______________________________________________ Gnustep-dev mailing list [email protected] https://lists.gnu.org/mailman/listinfo/gnustep-dev
