Re: [PATCH] gitattributes: Cover Objective-C source files

[Philippe Mathieu-Daudé]

On 19/3/22 04:25, Akihiko Odaki wrote:
> On 2022/03/19 1:14, Philippe Mathieu-Daudé wrote:
> > > Commit 29cf16db23 says:
> > > > Since commits 0979ed017f0 ("meson: rename .inc.h files to .h.inc")
> > > > and 139c1837db7 ("meson: rename included C source files to .c.inc")
> > > > 'git-diff --function-context' stopped displaying C function context
> > > > correctly.
> > >
> > > So I suspect Git has some knowledge of common file extensions like 
> > > .c, .h and .m although I couldn't find in the source code of Git.
> >
> > 'git-diff --function-context' doesn't work for me without this change.
>
> With some debugging, I found Apple's Git distribution actually carries a 
> default gitattributes file which annotates *.m.
> https://github.com/apple-opensource/Git/blob/master/gitattributes

I see, I'm using the Homebrew git. Anyway this change helps non-native
Darwin users to review these .m files.

> However, it does not annotate *.c or *.h. Apparently there is no "c" 
> diff pattern and they are handled with the "default" diff pattern which 
> is actually designed for C. In fact, "c" diff pattern is not present in 
> the documentation:
> https://git-scm.com/docs/gitattributes#_defining_an_external_diff_driver

'cpp' is listed. Maybe 'c' is aliased to it?

> In conclusion, *.m should be listed in gitattributes but *.c.inc and 
> *.h.inc should not be if my understanding is correct.

But then how git-tools can detect .inc are C files? I remember it was
not working (on Linux hosts) without this change.

> Paolo Bonzini, I found you are the author of commit 29cf16db23. Can you 
> test the above conclusion?
>
> Regards,
> Akihiko Odaki