Tests broken with native Windows perl

[MarcelRaad]

I did this

From an MSYS2 shell on Windows:

cd tests && ./runtests.pl

I expected the following

Successful test run. Instead, I get:

No such signal: SIGUSR1 at ./runtests.pl line 274.

This is with MSYS2's perl used in its MINGW32 shell:

$ perl --version
This is perl 5, version 32, subversion 1 (v5.32.1) built for MSWin32-x86-multi-thread-64int

Strangely, the error doesn't appear in the CI.

curl/libcurl version

2fe5241

operating system

Windows 10 (MSYS2)

CC @dfandrich

[dfandrich]

SIGUSR1 has been part of POSIX from the beginning. What signals are available in this perl?

[MarcelRaad]

perl -e 'print join(" ", keys %SIG)' says:

ABRT ALRM BREAK CHLD CLD CONT FPE HUP ILL INT KILL NUM05 NUM06 NUM07 NUM10 NUM12 NUM16 NUM17 NUM18 NUM19 NUM24 PIPE QUIT SEGV STOP TERM                

Maybe I also did something wrong, this is the first time I've set up an MSYS2 development environment in a long time, and the last time I've manually run the tests in MSYS2 was last year.

[dfandrich]

For comparison, SIGUSR1 is #10 on Linux, #16 on Solaris, #30 on FreeBSD and OpenBSD. "NUM10" and "NUM16" are both in your list, which makes it seem like those are placeholders for a signal that isn't supported on that OS or that perl version. The SIGUSR1 define has been a part of msys2 for at least 22 years according to git, but I don't know about perl. I've also just read that Windows doesn't even support sending signals like I'm using SIGUSR1 for, so maybe an error setting this signal handler should simply be ignored on Windows.

[MarcelRaad]

I've also just read that Windows
doesn't even support sending signals like I'm using SIGUSR1 for, so maybe an
error setting this signal handler should simply be ignored on Windows.

Sounds good to me!