Re: problem building Gnulib libgnu_a-malloca.o with Debian oldstable's clang

On 2026-08-06 G. Branden Robinson wrote:
> ../lib/malloca.c:49:56: error: too few arguments provided to function-like macro invocation
> static_assert (2 * sa_alignment_max - 1 <= (small_t) -1);
>                                                        ^
> /usr/include/x86_64-linux-gnu/sys/cdefs.h:447:10: note: macro '_Static_assert' defined here
> # define _Static_assert(expr, diagnostic) \
>          ^
> ../lib/malloca.c:49:57: error: expected '('
> static_assert (2 * sa_alignment_max - 1 <= (small_t) -1);

Based on the cdefs.h line number 447, this is on Debian Bullseye
(oldoldstable), not Debian Bookworm (oldstable). The old cdefs.h
doesn't check for __clang_major__ >= 4 which was added in glibc commit
c8ba52ab3350c3 ("misc: Sync cdefs.h with gnulib"), and then cdefs.h
defines _Static_assert as a macro when the compiler isn't in C11 or
newer mode. So the compiler probably is in C99 mode.

From groff-1.25.0.rc2/configure.ac:

# Autoconf [2.70,2.72] tries to cram C11 and C++11 down our throats.  We
# don't (yet) want them.
ac_cv_prog_cc_c11=no

I didn't investigate if the above is the reason why Gnulib doesn't
override static_assert with its own version that support C23's
single-argument form.

Maybe not relevant, but building groff on a modern distro with
CC='clang -std=gnu99' results in these lines in Makefile:

CC = clang -std=gnu99
CPP = clang -std=gnu99 -std=gnu11 -E

That is, even if ac_cv_prog_cc_c11=no is forced, CPP will be in GNU C11
mode.

-- 
Lasse Collin