Re: setlocale: Silence a -Wdiscarded-qualifiers warning on Windows.

Alejandro Colomar <[email protected]> Mon, 8 Jun 2026 01:54:49 +0200
Newsgroups	gmane.comp.lib.gnulib.bugs
Message-ID	<aiYBC-IqomJ96Pcy@devuan>
Hi Bruno, Collin,

On 2026-06-07T21:35:07+0200, Bruno Haible wrote:
> Collin Funk wrote:
> > > But we can add the intended return type in Gnulib. Done as follows. This
> > > patch removes two casts, instead of adding a cast.
> > 
> > I considered suggesting this as well for the same reason you mention,
> > i.e., the standard saying the returned string cannot be modified. But I
> > mistakenly assumed you would not want to diverge from the standardized
> > prototype.
> 
> We had some (conditional) change of return type from 'char *' to 'const char *',
> or from 'void *' to 'const void *', in functions like strchr and memchr,
> and although it was a bit of a hassle for the people who maintain very old
> packages, it was a worthwhile modernization.
> 
> In POSIX, the following functions are marked with
>   "The application shall not modify the string returned."
> and could therefore be modernized:
>   ctermid
>   dlerror
>   getgrent
>   getgrgid, getgrnam
>   gethostent
>   getlogin
>   getnetent, getnetbyaddr, getnetbyname
>   getprotoent, getprotobyname, getprotobynumber
>   getpwent, getpwnam, getpwuid
>   getservent, getservbyname, getservbyport
>   localeconv
>   nl_langinfo, nl_langinfo_l
>   ptsname
>   setlocale
>   strerror, strerror_l
>   strsignal
>   ttyname
>   *gettext, *gettext_l

Of which, the following are in ISO C:

	localeconv
	setlocale
	strerror

> > I vaguely remember
> > POSIX being willing to change the return type of gettext to be "const
> > char *" for similar reasons
> 
> That would be useful too, yes.
> 
> > Perhaps it is worth bringing this up with the ISO C people.
> 
> Yes, that would be useful, both for setlocale() and for localeconv().

I will mention within the committee that now that we've changed the
prototypes of string APIs --thus accepting breaking changes--, we could
embrace the const-correctness even further by changing the return types
of those functions too.

Thanks for the suggestion!

And indeed, I agree that changing the return values of those functions
is a good idea, as a gnulib extension (and hopefully also in the
standards).

I've reviewed 8724b4400aac (2026-06-07; "setlocale: Detect invalid
writes to the returned string in some cases."), and it LGTM.  Thanks!

Should I also have a look at anything else from this thread?


Have a lovely night!
Alex

> 
> Bruno

-- 
<https://www.alejandro-colomar.es>
signature.asc (application/pgp-signature, 833 B)