Issue deprecation warning when item parameter of pytest_warning_captured is used

[nicoddemus]

Follow up to #4036

[RonnyPfannschmidt]

@nicoddemus pluggy cant support this - we have to deprecate the complete hook and give it a new name

[nicoddemus]

You are right, not sure what I was thinking

[RonnyPfannschmidt]

@nicoddemus i propose pytest_warning_recorded(record) as a starting point, perhaps also adding a config parameter

[gnikonorov]

Hi @nicoddemus and @RonnyPfannschmidt is there an update on this? I tried to use pytest_warning_captured as detailed here and received the error in the attached log. When I tried to use pytest_warning_recorded(record) I received a PluginValidationError saying the hook name was invalid.

pytest_warning_captured_usage.log

[RonnyPfannschmidt]

@gnikonorov as far as I can tell nothing happened

[gnikonorov]

@RonnyPfannschmidt can I submit a PR for this then?

[RonnyPfannschmidt]

yes please

[gnikonorov]

Thank you! I'll try to have the PR up in the next few days

[gnikonorov]

Hi @RonnyPfannschmidt, I've made a PR for this, but I am a little confused about how to understand the codecov reports. Since I deprecated a hook and replaced it with a new one, I believe I'm maintaining the same test coverage/functionality as before but codecov says I've hit 90.90%. Is it also looking at hookspec.py and reference.rst?

[bluetech]

From the log you attached, the error seems to be something different than this issue:

INTERNALERROR> Traceback (most recent call last):
INTERNALERROR>   File "/home/gnikonorov/.local/share/virtualenvs/pytest-html-k1-ITgOR/lib/python3.6/site-packages/pluggy/callers.py", line 171, in _multicall
INTERNALERROR>     args = [caller_kwargs[argname] for argname in hook_impl.argnames]
INTERNALERROR>   File "/home/gnikonorov/.local/share/virtualenvs/pytest-html-k1-ITgOR/lib/python3.6/site-packages/pluggy/callers.py", line 171, in <listcomp>
INTERNALERROR>     args = [caller_kwargs[argname] for argname in hook_impl.argnames]
INTERNALERROR> KeyError: 'location'

It's because one of the call sites of the hook are missing the location parameter. I pointed this out in the review of #7255. As also mentioned there, pytest-xdist seems to already have a workaround for this issue. If you fix the location problem in pytest, does it work for you?

[gnikonorov]

Thanks @bluetech you are 100% right. I jumped the gun and assumed that resolving this would fix my issue too. Adding location=None resolves the problem. I still think we should add the new hook, but I will also fix this location bug in the new hook and set it to its proper value ( I'm not sure if None is the correct one, just a test value ).

What do you think @RonnyPfannschmidt?

Relevant log/diff: when_location_set_to_none.txt

[bluetech]

I will also fix this location bug in the new hook and set it to its proper value ( I'm not sure if None is the correct one, just a test value ).

Can you also fix it in the existing hook (in a separate PR)? (If you prefer, I can do it myself)

[gnikonorov]

Sure @bluetech, I'll investigate further and make a follow up issue/pr if we need to use a different value. I investigated for a while and None seemed fine which is why I used it for the PR associated with this issue.