trying to fix cygwin CI, cf #216

[dimpase]

trying to fix cygwin CI, cf #216

[dimpase]

Cygwin build still fails - it now can find mesonpy, but still fails pretty fast, and the log shows that meson mixes toolsets quite badly.

[dimpase]

ninja (invoked from meson?) picks up wrong Python headers - instead of Cygwin's Python, it takes native ones, oops:

Found ninja.EXE-1.12.0 at C:\tools\cygwin\bin\ninja.EXE
  + meson compile
  [1/3] Compiling Cython source 'D:/a/cysignals/cysignals/src/cysignals/signals.pyx'
  [2/3] Compiling C object src/cysignals/signals.cp39-win_amd64.pyd.p/meson-generated_src_cysignals_signals.pyx.c.obj
  FAILED: src/cysignals/signals.cp39-win_amd64.pyd.p/meson-generated_src_cysignals_signals.pyx.c.obj
  "gcc" "-Isrc\cysignals\signals.cp39-win_amd64.pyd.p" "-Isrc\cysignals" "-I..\src\cysignals" "-Isrc" "-I..\src" "-IC:\hostedtoolcache\windows\Python\3.9.13\x64\Include" "-fvisibility=hidden" "-fdiagnostics-color=always" "-DNDEBUG" "-D_FILE_OFFSET_BITS=64" "-Wall" "-Winvalid-pch" "-Wextra" "-Wpedantic" "-O3" "-DCYTHON_CLINE_IN_TRACEBACK=0" "-U_FORTIFY_SOURCE" "-DMS_WIN64=" -MD -MQ src/cysignals/signals.cp39-win_amd64.pyd.p/meson-generated_src_cysignals_signals.pyx.c.obj -MF "src\cysignals\signals.cp39-win_amd64.pyd.p\meson-generated_src_cysignals_signals.pyx.c.obj.d" -o src/cysignals/signals.cp39-win_amd64.pyd.p/meson-generated_src_cysignals_signals.pyx.c.obj "-c" src/cysignals/signals.cp39-win_amd64.pyd.p/src/cysignals/signals.pyx.c
  In file included from C:\hostedtoolcache\windows\Python\3.9.13\x64\Include/Python.h:50,
                   from src/cysignals/signals.cp39-win_amd64.pyd.p/src/cysignals/signals.pyx.c:16:
  C:\hostedtoolcache\windows\Python\3.9.13\x64\Include/pyport.h:741:2: error: #error "LONG_BIT definition appears wrong for platform (bad gcc/glibc config?)."
...

[tobiasdiez]

I've no idea why this is failing now - there were no changes to this workflow and also no updates to the cygwin action...very strange.

[dimpase]

maybe a change to the host environment?

[dimpase]

there is also an error related to chocolatey shown in the log, some kind of version mismatch.

[dimpase]

I've opened https://github.com/egor-tensin/setup-cygwin/issues/14

[dimpase]

I eliminated the chocolatey warning in the action update, but it hasn't changed anything. I'm giving up. Let's disable the cygwin CI, maybe?

[tornaria]

I eliminated the chocolatey warning in the action update, but it hasn't changed anything. I'm giving up. Let's disable the cygwin CI, maybe?

It's running python 3.9 which we shouldn't waste time supporting anyway.

In fact, I am needing some python 3.11 and maybe 3.12 to fix the sig_occurred() mishap (https://docs.python.org/3/c-api/exceptions.html#c.PyErr_GetHandledException and https://docs.python.org/3/c-api/exceptions.html#c.PyErr_GetRaisedException), but I'll try to workaround it for earlier python. For PyErr_GetRaisedException() one can use PyErr_Fetch(), but I'm not sure there's an alternative to PyErr_GetHandlerException() before python 3.11.

I assume dropping python 3.9 is ok, but I'm not sure about python 3.10.

[tornaria]

I assume dropping python 3.9 is ok, but I'm not sure about python 3.10.

https://scientific-python.org/specs/spec-0000/#2024---quarter-4 seems to suggest it might be ok to drop python 3.10 if necessary.

[dimpase]

I assume dropping python 3.9 is ok, but I'm not sure about python 3.10.

cygwin is stuck on 3.9, so this seems like dead wood.