The only purpose of this change is to avoid using clinic global variable in def bad_argument() method. Just for that, I had to modify many lines :-(

cc @AlexWaygood

Please give me some time to look at this! :-)

"Check if generated files are up to date (pull_request)" failure looks unrelated:

./configure --config-cache --with-pydebug --enable-shared
    PKG_CONFIG_PATH: /opt/hostedtoolcache/Python/3.11.6/x64/lib/pkgconfig

configure: loading cache config.cache
configure: error: `PKG_CONFIG_PATH' has changed since the previous run:
configure:   former value:  `/opt/hostedtoolcache/Python/3.11.5/x64/lib/pkgconfig'
configure:   current value: `/opt/hostedtoolcache/Python/3.11.6/x64/lib/pkgconfig'

I rebased my PR on the main branch to try to fix the "Check if generated files are up to date" job failure.

Please give me some time to look at this! :-)

Mmmmh, so do you need more time to review?

Well, since @AlexWaygood is not available for review, i think that i will just merge my change next days.

Thanks for waiting, sorry for the slow response from me! I wanted to take the time to think about this properly, but haven't had a chance recently due to being busy at work, and it's unlikely that I'll have the time in the next week either.

My instinct is that there should be a simpler solution here that involves fewer changes, but if you feel like there's a rush to get this merged for whatever reason, then please go ahead. I agree I've kept you waiting for a while :-)

me:

The only purpose of this change is to avoid using clinic global variable in def bad_argument() method. Just for that, I had to modify many lines :-(

@AlexWaygood:

My instinct is that there should be q simpler solution here that involves fewer changes,

Yeah, there should be a way to redesign the code to have to pass less arguments. I dislike the current API :-( Moreover, passing limited_capi and clinic is kind of redundant, since using clinic.limited_capi is the same as passing limited_capi.

I close my PR for now. We can revisit this code later when we will try to give rid of the global clinic variable.