make regen-all doesn't pass -f to argument clinic

[nineteendo]

Bug report

Bug description:

Output:

python3.13 ./Tools/clinic/clinic.py --make --exclude Lib/test/clinic.test.c --srcdir .
Error in file 'Modules/posixmodule.c' on line 13087:
Checksum mismatch! Expected '9756767bdbdabe94', computed '1da32369b7d9fd4b'. Suggested fix: remove all generated code including the end marker, or use the '-f' option.
make: *** [clinic] Error 1

This makes it harder to fix merge conflicts, see: https://discuss.python.org/t/reducing-merge-conflicts-in-generated-code/68537/8

CPython versions tested on:

CPython main branch

Operating systems tested on:

macOS

Linked PRs

• gh-126206: make clinic now forcefully regenerates clinic code #126244

[erlend-aasland]

If you get conflicts in generated code you should examine the conflict. If there are real issues, fix them; if not, regenerate clinic code using -f, as suggested by the error message. See also the devguide and the Argument Clinic HOWTO.

[nineteendo]

Sure, if the generated code is in .c files should probably be examined as it might contain incompatible changes. But why can't it overwrite .c.h files automatically? That code is fully generated (at least for Modules/posixmodule.c).

I know that I can run Tools/clinic/clinic.py -f Modules/posixmodule.c, but this just a waste of time.

[erlend-aasland]

There is no discussion of using -f in the original issue nor in the commit message that added the clinic target. I'm not sure why we don't forcefully recreate generated code, but I would be fine with such a change. The clinic-tests target, for example, uses the -f option when running clinic.py.

[erlend-aasland]

Adding -f would not conflict with the check-generated-files CI check, so we should be fine CI wise.

[erlend-aasland]

Thanks for the suggestion; make clinic (and any other target that implicitly depends on clinic) now forcefully regenerates clinic code.