Version bump CI tools

[felixdivo]

The automated static analysis tools are updated to their most recent version.

EDIT: I basically updated the tools in requirements-lint.txt and some images in Travis CI. Then I and re-ran the formatter, which causes much of the diff. I also removed the formatting job from Travis as it was already done by Github actions. I removed the normal testing from Travis, as it is much easier to specify in Github actions (and we now also test more platforms, like properly test on macOS!). We now also test on newer versions of Python as well (currently up to 3.10.0-alpha.4) and on Pypy on all platforms (also new).

[hardbyte]

🎖️

[felixdivo]

How useful is it to have both Github actions as well as AppVeyor an Travis CI? Wouldn't it be nice to at least remove the tests from Travis CI as they seem to be very slow and a duplicate. Also, do we need AppVeyor at all?

EDIT: Proposal:

• retain all Github workflow tests, which test with CPython on Linux, macOS and Windows
• Move CPython 3.10-dev & nightly from Travis to Github workflow
• remove Travis CI testing except: pypy3 and socketcan tests
• retain the other Travis jobs like the linter and publishing (for now at least, we can consider moving them in another PR)
• remove AppVeyor testing as it is entirely covered by Github workflows (right?)

[hardbyte]

Travis-CI might be a bit faster now I just migrated to the travis-ci.com version.

But yes I think just relying on github workflow except where we need additional functionality is the way forward.

Is there more you want to do in this PR?

[felixdivo]

I basically updated the tools in requirements-lint.txt and now also in .travis.yml. I'd wait for the discussion in #951 before doing more, since the macOS build seems to fail for now but there is no need for debugging if we delete anyways.

[felixdivo]

Seems like this uncovered an actual problem: https://github.com/hardbyte/python-can/pull/940/checks?check_run_id=1458088551#step:4:101. Pypy support for that lib is aparrently not available, see python/typed_ast#111.

[codecov]

Codecov Report

Merging #940 (316a881) into develop (a39949f) will decrease coverage by 0.75%.
The diff coverage is 27.77%.

@@             Coverage Diff             @@
##           develop     #940      +/-   ##
===========================================
- Coverage    70.28%   69.53%   -0.76%     
===========================================
  Files           76       76              
  Lines         7398     7455      +57     
===========================================
- Hits          5200     5184      -16     
- Misses        2198     2271      +73     

[felixdivo]

I also opened ionelmc/python-lazy-object-proxy#46

EDIT: was redirected to pypa/wheel#382

[felixdivo]

The problem on Python version 3.10 with python-lazy-object-proxy should by fixed by wheel version 0.36.2.

[felixdivo]

Seems like this time, we actually uncovered a bug in our repository: https://github.com/hardbyte/python-can/pull/940/checks?check_run_id=1753568353#step:5:333

[zariiii9003]

We could set python 3.10 to experimental and merge this. What do you think?

https://docs.github.com/en/actions/reference/workflow-syntax-for-github-actions#jobsjob_idcontinue-on-error

[zariiii9003]

Here's a new one (for me): https://github.com/hardbyte/python-can/pull/940/checks?check_run_id=2336410553#step:5:332

And the pypy3 threading error also occurs on windows.

[felixdivo]

Here's a new one (for me): https://github.com/hardbyte/python-can/pull/940/checks?check_run_id=2336410553#step:5:332

Should be fixed. The runners have such an unreliable timing that any kind of deadline/timing assert seems to fail eventually.

[felixdivo]

Asyncio error in python 3.10 is tracked in #1005

[felixdivo]

@zariiii9003 Everything seems to be done now, right?

[zariiii9003]

Depends, do we keep Appveyor?

[felixdivo]

It would also touch a few (even more) lines, I'd just do it in a separate PR right after this one is merged. It's already somewhat too large I think.

[felixdivo]

@hardbyte Do you wish to do/have someone do another review or is it OK to be merged as is?

[hardbyte]

Good work hunting down all the loose ends! Happy to merge

[felixdivo]

@zariiii9003 Nice work, thanks for joining. I'll assemble a much simpler PR that will remove AppVeyor in a minute.