Fix make error for Fedora 42

[michaelortmann]

Found by: robert-scheck
Patch by: michaelortmann
Fixes: #1740

One-line summary:
Add -std=gnu99, if available, to CFLAGS

Additional description (if needed):
GCC 15 changes the default language version for C compilation from -std=gnu17 to -std=gnu23. See https://gcc.gnu.org/gcc-15/changes.html

Please run misc/runautotools when merging this PR.

Test cases demonstrating functionality (if applicable):
Test with gcc-15-20250112
Before:

/home/michael/opt/gcc-15-20250112/bin/gcc -g -O2 -pipe -Wall -I.. -I..  -DHAVE_CONFIG_H -I/usr/include  -c botcmd.c
botcmd.c: In function ‘bot_filereq’:
botcmd.c:1155:9: error: too many arguments to function ‘f’; expected 0, have 3
 1155 |         f(idx, from, path);
      |         ^ ~~~
make[1]: *** [Makefile:86: botcmd.o] Error 1
make[1]: Leaving directory '/home/michael/projects/eggdrop/src'
make: *** [Makefile:243: eggdrop] Error 2

After:
/home/michael/opt/gcc-15-20250112/bin/gcc -std=gnu99 -g -O2 -pipe -Wall -I.. -I.. -DHAVE_CONFIG_H -I/usr/include -c botcmd.c

[robert-scheck]

Wouldn't the real fix have been to correct the prototype of f() instead of adding a compiler flag to suppress this?

[thommey]

We rely on the f() syntax to specify a function with unspecified arguments. That is perfectly legal until and including C17. Eggdrop in general follows POSIX 2001 and C99 with GNU extensions. I believe it is the correct solution to specify what coding standard we use on the commandline (gnu99) and I was surprised we don't yet do that.
Otherwise, like in this instance, compilers changing their default can break our code if the standards aren't compatible, we don't want that.

I am confused by the attempt to make C23 saner by introducing f() = f(void) so late in the game, I would prefer to avoid C over these breaking language changes.

Update: The justification for the change was given as (https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2841.htm):

Function declarators without prototypes have been an obsolescent feature since the very first ISO C standard. 33 years of advance warning should be enough.

Hard to argue with that, I was under the impression it was legal and supported. In any case, we would have to rewrite a massive portion of our codebase (as we use f() as a generic function prototype for the module API) - @michaelortmann started with that at first, but that is simply too much change in a legacy codebase with little test coverage IMHO.