bug#42538: 28.0.50; tramp-test35-remote-path test timing out on macOS

[Philipp Stephani]

Am Fr., 31. Juli 2020 um 20:04 Uhr schrieb Michael Albinus
<michael.albinus@gmx.de>:
>
> Philipp Stephani <p.stephani2@gmail.com> writes:
>
> Hi Philipp,
>
> >> Thanks for the analysis. However, I fail to see where Tramp fires the
> >> command 'kill -17 $$'. There's only 'kill -2 $$'. Could you, please, show 
> >> me?
> >
> > `tramp-get-signal-strings' invokes "bash -c 'kill -NUM $$'" for all
> > signal numbers, including 17. It excluded 19, assuming that's SIGSTOP,
> > but that assumption doesn't hold on macOS (or FreeBSD, or any non-x86
> > system, FWIW).
>
> I see. `tramp-get-signal-strings' is problematic anyway, because it runs
> locally, but shall serve the signal strings for the remote host. Maybe,
> we shall assume standardized signal strings up to 15 only.

Even that assumption doesn't hold, even if restricted to a single
operating system. See e.g. the per-architecture signal number table at
https://man7.org/linux/man-pages/man7/signal.7.html.
How about using the fallback ("Signal #123") in all cases? In practice
the signal descriptions aren't portable, so users can't parse them
anyway.