Upgrade to aioredis 2

[Dreamsorcerer]

Fixes #528.

[Dreamsorcerer]

@webknjaz Do you have any idea what the error in CI means? I tried removing the loop() fixture in conftest.py in the hope of fixing other issues, but now it's complaining about a loop() fixture in aiohttp's plugin...

Specific changes that introduced this error: https://github.com/aio-libs/aiohttp-session/pull/611/files/9bf3ff0e3e8f2800c5dc0d39525f106009ad1ca5..5f0ffbd7932e2ee256e0fb0ae4cdcad50d73cddc

[codecov]

Codecov Report

Merging #611 (c85be12) into master (4de3db4) will increase coverage by 0.82%.
The diff coverage is 75.00%.

@@            Coverage Diff             @@
##           master     #611      +/-   ##
==========================================
+ Coverage   96.22%   97.04%   +0.82%     
==========================================
  Files           6        6              
  Lines         344      338       -6     
  Branches       43       42       -1     
==========================================
- Hits          331      328       -3     
+ Misses          7        5       -2     
+ Partials        6        5       -1     

Flag	Coverage Δ
unit	97.04% <75.00%> (+0.82%)	⬆️

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

Impacted Files	Coverage Δ
aiohttp_session/redis_storage.py	97.72% <75.00%> (+5.72%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e3c4eea...c85be12. Read the comment docs.

[Dreamsorcerer]

@webknjaz Can we make codecov/patch optional? This is the 2nd time it's blocked me. It's complaining about 3 lines of untested code which have only had comments/annotations changed.

[webknjaz]

@webknjaz Can we make codecov/patch optional?

How about patching codecov.yml to be less sensitive to small changes?

[Dreamsorcerer]

@webknjaz Can we make codecov/patch optional?

How about patching codecov.yml to be less sensitive to small changes?

I don't see anything in the documentation that might do this. It's also not really the size, but the fact that it doesn't know the difference between functional changes and style/comment changes. It just seems rather awkward trying to hack the config to convince codecov to pass. Code coverage is useful as a guiding tool, I don't see it being useful as a hard rule that can't be bypassed when we know better (and we don't require it on the aiohttp repo...).

[Dreamsorcerer]

On aiohttp you enabled an option to skip the chronographer with a label. If that's easy to setup for codecov as well, I'd be happy with that.

[webknjaz]

On aiohttp you enabled an option to skip the chronographer with a label. If that's easy to setup for codecov as well, I'd be happy with that.

Chronographer is an app that I wrote so I implemented this in its code. We don't have such a level of control over Codecov. But I'm pretty sure it's possible to make it less sensitive.

[webknjaz]

@Dreamsorcerer you probably need to adjust this https://docs.codecov.com/docs/commit-status#branches

[Dreamsorcerer]

@Dreamsorcerer you probably need to adjust this https://docs.codecov.com/docs/commit-status#branches

The branches option? Wouldn't that just result in it not reporting anything for some branches? That would make it impossible to merge as it is a required check.

There are other options there like informational which would just make the check pass regardless, but that's fundamentally the same as removing the required status from Github, except it would make it harder for maintainers to notice when it should be failing.

[webknjaz]

Ah, sorry, I somehow copied the wrong link. Take a look at the patch status options: https://docs.codecov.com/docs/commit-status#patch-status.

[Dreamsorcerer]

Ah, sorry, I somehow copied the wrong link. Take a look at the patch status options: https://docs.codecov.com/docs/commit-status#patch-status.

I still don't see anything useful there...
I also think (unless they learn to filter out cosmetic changes) that this should have failed, and therefore alerted the maintainer. The maintainer can then make an informed decision about whether there needs to be a new test or not.

[webknjaz]

I don't think so. If a patch adds uncovered lines, it should catch this. I think that setting a threshold of 75% would make it possible.

[webknjaz]

FWIW I dropped the patch requirement for now.