Ignore RedisConnectionError and exclude TypeChecking blocks from coverage

[armanddidierjean]

No description provided.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 81.04%. Comparing base (08dcf2b) to head (8cd8904).
Report is 11 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #620      +/-   ##
==========================================
- Coverage   81.26%   81.04%   -0.22%     
==========================================
  Files         128      130       +2     
  Lines        9949    10100     +151     
==========================================
+ Hits         8085     8186     +101     
- Misses       1864     1914      +50     

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

[Daihecyy]

Consider changing xfail to skipif since the tests seem to be expected to run in Github actions.
Maybe create a shareable skipif decorator in test_commons ?

[armanddidierjean]

Consider changing xfail to skipif since the tests seem to be expected to run in Github actions. Maybe create a shareable skipif decorator in test_commons ?

I'm not sure how I can decide if the test should be run as settings are not loaded when the file is imported. I think that failed test should raise an error in github action

[Daihecyy]

Consider changing xfail to skipif since the tests seem to be expected to run in Github actions. Maybe create a shareable skipif decorator in test_commons ?

I'm not sure how I can decide if the test should be run as settings are not loaded when the file is imported. I think that failed test should raise an error in github action

But the tests passing would be considered "unexpedly passing" XPASS which is not really what we want.
We could do a similar thing as with override_get_db in tests/commons and return None if the redis client is not configured.
We could then skipif with the condition the redis client being None.
Such an approach would be a lot cleaner as we would not allow our tests to fail.

[Daihecyy]

I must say that the way redis_client is handled in tests/commons is really weird to me.
Such a type redis.Redis | None | bool seems to indicate something is wrong.