> The error message from the main process is a consequence of the error
> in the child process. The comment says that apparently, non-blocking sockets
> don't work. The code in gnulib/lib/nonblocking.c uses Windows system calls
> with pipes.
>
> Maybe the cause is that in your CI you use
>
> defaults:
> run:
> shell: C:\cygwin\bin\bash.exe --login -e {0}
>
> where my CI has
>
> defaults:
> run:
> shell: C:\cygwin\bin\bash.exe -eo pipefail -o igncr '{0}'
>
> ?
I first tried starting bash with "-o errexit" [1], but
test-nonblocking-socket.sh still failed [2].
Then I tried with "-o errexit -o pipefail" [3], and
test-nonblocking-socket.sh failed again [4].
The error is always the same:
FAIL: test-nonblocking-socket.sh
================================
../../gltests/test-nonblocking-reader.h:111: assertion 'spent_time < 0.5' failed
test-nonblocking-socket-main.exe: ./test-nonblocking-socket-child.exe
subprocess got fatal signal 15
../../gltests/test-nonblocking-socket-main.c:119: assertion 'err == 0' failed
FAIL test-nonblocking-socket.sh (exit status: 1)
--
Michele
[1]
https://github.com/mlocati/gnulib-cygwin-tests/commit/a7baa4976d89d03f5277eb1c8a472f61a9b035b7
[2] https://github.com/mlocati/gnulib-cygwin-tests/actions/runs/11068572586
[3]
https://github.com/mlocati/gnulib-cygwin-tests/commit/fa4ce13b1612901b6ec3a37f9726aae340379cc1
[4] https://github.com/mlocati/gnulib-cygwin-tests/actions/runs/11068658216
