Re: autoconf-2.73 and gcc-14 and incomplete C23 support

Paul Eggert <[email protected]> Wed, 29 Apr 2026 00:30:21 -0700
Newsgroups gmane.comp.sysutils.autoconf.bugs
Organization UCLA Computer Science Department
Message-ID <[email protected]>
On 2026-04-28 15:07, Karl Berry wrote:

> I think it would be better if autoconf did not switch to such partial
> implementations of standards, and only switched if the compiler supports
> the full, final, real standard.

Unfortunately no compiler does that, as far as I know, and even if it 
did it would be hard to test for it.


> P.S. Also, not a bug, but I admit I don't understand the logic of
> preferring C23 for C compilers by default (when available), but leaving
> C++ as the compiler default.

To some extent this is a historical accident. C compilers historically 
have been verrrrry conservative: they have been quite laggy about 
default standards, and choosing a different default typically has not 
broken libraries or apps and has been helpful overall. This is less true 
for C23 than it was for C17, C11, and C99 but it's still pretty true.

C++ has been different: a C++ compiler+standard library typically does 
not work if you choose a standard newer than (or often, even older than) 
the default.

All that being said, it would be nice to have a documented-way to say 
"Don't change the C compiler's standard version; just use the default."