Re: Clang __has_feature check for new 'nullable' language extensions?
Re: Clang __has_feature check for new 'nullable' language extensions?
- Subject: Re: Clang __has_feature check for new 'nullable' language extensions?
- From: Chris Lattner <email@hidden>
- Date: Thu, 12 Feb 2015 10:59:35 -0800
On Feb 12, 2015, at 9:18 AM, Jens Alfke < email@hidden> wrote:
On Feb 11, 2015, at 11:55 PM, Roland King < email@hidden> wrote:
I’m guessing it’s
__has_feature( nullability )
That works — thanks!
So the magic invocation I’m putting in my header files (after the #includes) is:
#if __has_feature(nullability) // Xcode 6.3+
#pragma clang assume_nonnull begin
#else
#define nullable
#define __nullable
#endif
… interface declarations go here…
#if __has_feature(nullability)
#pragma clang assume_nonnull end
#endif
In property declarations you'll need to make “nullable” the last meta-attribute, otherwise you’ll get syntax errors in earlier versions of Xcode; so for example “(readonly, nonatomic, nullable)” works but “(readonly, nullable, nonatomic)” is a syntax error because in Xcode 6.1 the parser will see two adjacent commas.
Hi Jens,
FYI, you can use _Pragma (instead of #pragma) to wrap this up in a nicer macro, but we’re trying to get this taken care for you in a future Beta (something along the lines of NS_ASSUME_NONNULL_BEGIN/END).
[This is working well, but in converting my interfaces I found a nasty compiler bug that I’m about to submit to Apple — if you declare multiple properties on a single line, adding the (nullable) attribute will cause the parser to lock up in an infinite loop: @property (readonly, nullable) id a, b, c; // don’t do this! I spent an hour or so with my MBP’s fan running full speed before figuring out which exact line triggered this.]
We’ll investigate, thanks!
-Chris
|
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden