- User Since
- Jan 29 2021, 10:57 AM (68 w, 2 d)
Thu, May 12
Mon, May 9
Jan 5 2022
I think that this is a good warning and I'll defer to the experts for what has to happen when prototypes merge with K&R definitions :)
Nov 12 2021
Nov 10 2021
Oct 29 2021
Thanks Arthur for your feedback.
Thanks for looking, Aaron. You're right that the main utility of the aggregation of format warnings is to extend C's type checking because there is no other good way, or good place, to do it. I have built hundreds of millions of shipping lines of C, C++ and Objective-C, and this change seems like it would be an effective fix in several places where we don't currently have anywhere else to go.
Oct 28 2021
Add test for a protocol method with format_arg, second NSString method accepting a NSString instead of a C string
Apologies, Phabricator showed the comment below line 197 in the diff, but the email showed it to be below line 3404. I can check if the return type is instancetype in handleFormatArgAttr and use that instead.
Forgot to run clang-format.
Oct 27 2021
Oct 26 2021
Thanks Artem for pointing out that I was completely misusing getFunctionScopeIndex. This should be better. I added a test that you can pick a non-1 value for the format parameter in blocks.
Feb 5 2021
Feb 3 2021
Feb 2 2021
Address Aaron's feedback
Jan 29 2021
Updating the diff using arcanist, which I'm told produces better results. Sorry for the churn!