ubuntu-22.04 fix: NoneType object has no attribute getcurrent()

[farshan-dev]

related issue: #953

[4383]

What do you think of making id(gr) the default return value rather than try/except/if/else?

>>> gr = None
>>> id(gr)
140586328307168

Calling id(None) seems to always return something so... why not.

The idea behind these changes LGTM, I'd just suggest to refactor the logic to reduce the complexity of this function. Don't want to pick nit neither. Let me know if that sounds good for you.

https://docs.python.org/3/library/functions.html#id

[4383]

On the other hand I wonder if by forcing the return an identifier even when gr is equal to None we could may find us hidding these symptoms. Symptoms which may be the manifestation of a more pernicious problem.

Please can you give us more details about the context where you face this problem?
Do you have a reproducer?

[4383]

if greenlet.getcurrent doesn't exist and if gr is equal to None seems weird.

[4383]

That's more or less the scenario you propose to follow.

[4383]

As this eventlet module (eventlet.green.thread) is the patched version of the Python thread module I'd argue that maybe, one option, could be to return the result of the native get_ident function. It will return the ‘thread identifier’ of the current thread. IMO, That would more meaningful that returning an integer computed from a None value.

https://docs.python.org/3/library/threading.html#threading.get_ident

But I still continue to think that the observed behavior is the symptom of a problem more deeply rooted.

[farshan-dev]

thank you for your review, after searching for root of this problem, I found its not related to this package, and its about bobcat keystone package that uses this package. and I tried to fix actual problem and eventlet package works fine.

for reproduce this package install openstack prereqs and set bobcat, then install keystone, and run su -s /bin/sh -c "keystone-manage db_sync", in ubuntu 22.04, you can see error.
thank you again 🙌

[4383]

Thanks for your feedback. So, if you agree with abandoning this proposal I think we can close this github pull request and its associated github issue. Do you agree?

[codecov]

Codecov Report

Attention: Patch coverage is 80.00000% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 56%. Comparing base (c22b896) to head (62f1e79).

Files	Patch %	Lines
eventlet/green/thread.py	80%	1 Missing ⚠️

Additional details and impacted files

@@          Coverage Diff          @@
##           master   #954   +/-   ##
=====================================
- Coverage      56%    56%   -1%     
=====================================
  Files          89     89           
  Lines        9767   9770    +3     
  Branches     1819   1819           
=====================================
- Hits         5477   5476    -1     
- Misses       3920   3923    +3     
- Partials      370    371    +1     

Flag	Coverage Δ
ipv6	23% <40%> (+<1%)	⬆️
py310asyncio	52% <80%> (-1%)	⬇️
py310epolls	53% <80%> (-1%)	⬇️
py310poll	53% <80%> (-1%)	⬇️
py310selects	53% <80%> (-1%)	⬇️
py311asyncio	52% <80%> (-1%)	⬇️
py311epolls	53% <80%> (-1%)	⬇️
py312asyncio	50% <80%> (-1%)	⬇️
py312epolls	51% <80%> (-1%)	⬇️
py37asyncio	50% <80%> (+<1%)	⬆️
py37epolls	51% <80%> (-1%)	⬇️
py38asyncio	51% <80%> (-1%)	⬇️
py38epolls	53% <80%> (-1%)	⬇️
py38openssl	51% <80%> (-1%)	⬇️
py38poll	53% <80%> (-1%)	⬇️
py38selects	53% <80%> (+<1%)	⬆️
py39asyncio	51% <80%> (-1%)	⬇️
py39dnspython1	51% <80%> (-1%)	⬇️
py39epolls	53% <80%> (-1%)	⬇️
py39poll	53% <80%> (-1%)	⬇️
py39selects	52% <80%> (-1%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.