Get mypy working again

[gaffney2010]

Resolves #1406

[gaffney2010]

Need to test adding new exception handling. Will reopen.

[drvinceknight]

Looks like this is failing on coverage:

axelrod/load_data_.py                                               37      1    97%   30

[gaffney2010]

Looks like this is failing on coverage:

axelrod/load_data_.py                                               37      1    97%   30

But why? That line is tested. Any idea?

[drvinceknight]

But why? That line is tested. Any idea?

Could it be the assertRaisesRegex doing something funky? Would it be worth trying assertRaises? That's a very underwhelming idea that I'm not convinced by myself.

[gaffney2010]

Could it be the assertRaisesRegex doing something funky? Would it be worth trying assertRaises? That's a very underwhelming idea that I'm not convinced by myself.

I actually had assertRaises that I changed thinking perhaps it wanted to test the text of the error message.

[gaffney2010]

Ah, so get_data raises FileNotFoundError in my test case, but returns None for some other kinda failure. I'll figure it out.

[gaffney2010]

I chose to inject the dependency on get_data for the sake of testing because:

1. It's a little hard to figure out this failure path
2. It seems to be missing tests even in the core library.
3. It may require the library to be formatted different (whatever PEP 302 is)

[gaffney2010]

Oh, mypy is giving a lot more errors. I'm not sure why I thought it was resolved. Will reopen after addressing the others.

[gaffney2010]

Oh, mypy is giving a lot more errors. I'm not sure why I thought it was resolved. Will reopen after addressing the others.

Ah, this was passing locally. I'm gonna continue to test via CI / build

[gaffney2010]

Everything seems to be passing now.

[drvinceknight]

I can't seem to get any information from read the docs about the failed build there.

Do the docs build locally for you? If they do it could be a read the docs thing instead of a sphinx thing...

[gaffney2010]

Do the docs build locally for you? If they do it could be a read the docs thing instead of a sphinx thing...

python doctest.py runs without errors

Is there a different command to build?

[marcharper]

That's just the doc tests, you need to also build the docs by running make. I thought we had this documented but I wasn't easily able to find it just now.

[drvinceknight]

That's just the doc tests, you need to also build the docs by running make. I thought we had this documented but I wasn't easily able to found it just now.

You're right Marc, that's not documented at all. We should put it here right? https://axelrod.readthedocs.io/en/stable/how-to/contributing/running_tests.html#testing-the-documentation.

I believe:

cd docs
make html

is what is needed.

[gaffney2010]

make html runs locally and creates no new files.

When we update the docs for make html, can we specify which version of Sphinx we're using - I had compatibility issues.

[marcharper]

@drvinceknight yes that seems like a good place.

Where are the commands that CI is running to build the docs?

[drvinceknight]

It's using the readthedocs config file which is here: https://github.com/Axelrod-Python/Axelrod/blob/dev/.readthedocs.yaml

Here's info about the file itself: https://docs.readthedocs.io/en/stable/config-file/v2.html

The rest is essentially done through the readthedocs configuration panel. As this isn't changing anything relevant to the docs how about we merge it, then I'll open a PR to fix the docs build?

[drvinceknight]

The fact that the other recent PRs pass rtd but not this one is weird...