Maintain aioresponses

[Polandia94]

I think aioresponses is the most used tool to mock aiohttp. however it looks unmaintained. Are you open to maintain a fork of aioresponses, or maybe to request the transfer of aioresponses to this org if the current maintainers accept?

I left a comment on aioresponses about this:
pnuckowski/aioresponses#281

And I created a repo with changes that i would like to see on aioresponses. I could maintain the fork there, but probablly is better to do it in some org like this (i will also ask on pytest)
https://github.com/Polandia94/aioresponses

[Dreamsorcerer]

I think it's a possibility, as we've seen a fair few users mentioning the library.

We do normally recommend sending actual requests though, rather than mocking like this. So, I'm not sure any aiohttp maintainers have used this library before.

[webknjaz]

I think this might create an additional burden on the current maintainers but if someone else were to maintain it, hosting it in the org could be acceptable. Though, I wonder if the aiohttp docs could be improved to shop better ways of doing testing so that the users wouldn't assume that lib is the way to go...

---

xref: pytest-dev/pytest#14251

[webknjaz]

@Polandia94 don't build the dists in the publishing job in CI. Don't give OIDC privileges to the build machinery. It's insecure. Build in a separate job, then publish. Best use the pypi-publish action for the latter so you'd have PEP 740 attestations out of the box.

[webknjaz]

The dependency declarations in pyproject.toml also use invalid syntax.

[webknjaz]

Don't add an arg to --cov, configure in the coveragepy config.

[Dreamsorcerer]

At a quick look, that seems like a good approach.
I'd suggest that mock_external_urls should just default to False, and that the initial README examples should all follow that pattern.
Maybe also have a small section explicitly for users migrating from aioresponses, which shows the minimal changes needed to get their tests working on this library.

I'd probably be fine with that coming under aio-libs then.

[Polandia94]

Thanks you for your recommendations. I changed the package with that and added a pytest fixture to keep only one server between a lot of tests.
I created PRs on some public repos to demostrate thow to change from aioresponses to aiointercept. I believe the package is ready, I need to improve the docs (probablly a readthedocs page) and the error messages (for example to check similar routes and methods when there is no match to get some message).
If you want I can transfer the ownership of repo and pypy to aio-libs or keep it under my account. let me know if you want to discuss it here or in another thread/space.