misleading error message

[Vincent Lefevre]

On 2024-09-24 09:20:51 +0200, Sergey B Kirpichev wrote:
> Hello,
> I believe, there is issue in the GMP, coming from changeset
> changeset:   18292:0f5102b5dea7
> user:        Marco Bodrato <bodrato at mail.dm.unipi.it>
> date:        Sun Dec 12 14:24:10 2021 +0100
> summary:     Handle overflow in mpz_type through errno.
> 
> included in 6.3.0 release.  Previously, for various cases in
> mpz/init2.c, "gmp: overflow in mpz type" was printed to stderr and
> abort() called.  Now GMP instead just calls  __gmp_exception(), that
> raise SIGFPE and abort().

It will not do both in general. The "raise (SIGFPE);" will normally
terminate the process with a core dump. The abort() will be called
only if SIGFPE is ignored, but it is dangerous to ignore this signal
as this may yield undefined behavior if it is generated by another
cause: in the signal(2) man page:

    According to POSIX, the behavior of a process is undefined
    after it ignores a SIGFPE, SIGILL, or SIGSEGV signal that was
    not generated by kill(2) or raise(3). Integer division by zero
    has undefined result. On some architectures it will generate a
    SIGFPE signal. (Also dividing the most negative integer by -1
    may generate SIGFPE.) Ignoring this signal might lead to an
    endless loop.

> System prints something like "Floating point exception", which might
> look misleading, as the program termination actually related to
> memory management.

I agree that this is misleading.

[...]
> Now I also see "Starting in early 2024, we cannot deliver email to
> any gmail.com addresses." I suspect it's because missing DMARC
> policy for gmplib.org. Why not set SPF & DMARC for this domain? This
> should fix your problems with Gmail (and, perhaps, with more and
> more other email providers, which adopt similar policies).

This is a stupid requirement from Gmail, in particular for users
of multiple machines and/or mailing-lists (one has to use "all"
for SPF, which is misleading).

-- 
Vincent Lefèvre <vincent at vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)