Re: [PATCH] fix incomplete declaration for C23

[Georg Sauthoff]

Hello,

On Sun, Jan 26, 2025 at 01:07:59AM +0000, Tim Rice wrote:
> > Just to add some more context, ftr, without that change datamash
> > compilation runs into a compile error on C23 compilers because C23
> > basically made `T foo()` equivalent to `T foo(void)`.
> 
> Yep, I believe that's what the commit message now says. Did I make mistakes?

I hadn't checked your commit message.

After revisiting the thread I just thought providing some additional context
directly in this mailing list thread might be useful to the casual reader - or
even myself in case I come across it after several months or so.

> > This broke the build on current Fedora Rawhide, recently, since the GCC
> > version available there already defaults to compiling to C23 (or to
> > C23+GNU extensions or so).

> Thanks for the extra context. I'm on gentoo and saw no problems here until I 
> explicitly set std.
 
> $ gcc -v
[..]
> gcc version 14.2.1 20241221 (Gentoo 14.2.1_p20241221 p7) ```
 
> Is it a GCC version thing, or is it a distro configuration thing?

It's the former.

Fedora Rawhide already is on GCC 15:

```
$ rpm -q gcc
gcc-15.0.1-0.3.fc42.1.x86_64
$ man gcc 2>/dev/null | grep ' gnu23' -A3
           gnu23
           gnu2x
               GNU dialect of ISO C23.  This is the default for C code.  The 
name gnu2x is deprecated.

$ gcc -dM -E -x c /dev/null | grep STDC_V
#define __STDC_VERSION__ 202311L
```


Best regards,
Georg

-- 
'Only bad boys use such recursive calls, but only good girls use
 this package.  Thus the problem is of minor interest.'
    (Listings user guide)