Re: [GIT PULL] __auto_type conversion for v6.19-rc1

On December 8, 2025 7:22:06 PM PST, Al Viro <[email protected]> wrote:
>On Mon, Dec 08, 2025 at 04:28:11PM -0800, H. Peter Anvin wrote:
>> On December 8, 2025 4:25:19 PM PST, Al Viro <[email protected]> wrote:
>> >On Mon, Dec 08, 2025 at 03:55:26PM -0800, H. Peter Anvin wrote:
>> >> Hi Linus,
>> >> 
>> >> The following changes since commit c2f2b01b74be8b40a2173372bcd770723f87e7b2:
>> >> 
>> >>   Merge tag 'i3c/for-6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/i3c/linux (2025-12-08 11:25:14 +0900)
>> >> 
>> >> are available in the Git repository at:
>> >> 
>> >>   git://git.kernel.org/pub/scm/linux/kernel/git/hpa/linux-auto.git
>> >> 
>> >> for you to fetch changes up to branch auto-type-for-6.19
>> >> (4ecc26fa585216f98d71411ce182f9e823d94c8c):
>> >> 
>> >>   tools/virtio: replace "__auto_type" with "auto" (2025-12-08 15:32:15 -0800)
>> >
>> >Argh...  teaching declaration parser in sparse to handle that is
>> >going to be fun, especially since there are corner cases where
>> >gcc and clang do not agree, even with --std=c23 --pedantic...
>> 
>> Well, until sparse actually handles C23, this is just a macro.  __auto_type is already in use.
>
>Just anticipating the joy of getting declaration parser to deal with that
>properly - there's bunch of fun corner cases where this macro wouldn't
>cut it.  Sure, the underlying semantics can be mapped onto __auto_type,
>but the actual syntax is bloody awful, especially when you mix the
>typedefs into it.
>
>Speaking of other fun sparse stuff: __VA_OPT__ support needs to be added;
>I think I have it plotted down to reasonable details, will post in a day
>or two...

Yeah... the C committee even admitted they botched the spec; the intent was for it to work "exactly like gcc __auto_type"...