Avoid requests automatic auth

[ryanhiebert]

Fixes #278

I haven't added any tests, but I can later.

[Lukasa]

So, this definitely looks plausible. I think we need some testing: in particular, I think we need some testing about how this interacts with the other places we process auth (see anywhere that has auth= in the codebase).

[ryanhiebert]

I'll try to put in some more time this evening to figuring out the testing.

[singingwolfboy]

Hey @ryanhiebert, what's the status of this pull request?

[coveralls]

Coverage increased (+0.03%) to 88.05% when pulling 0c9287f on ryanhiebert:noop-auth into 4fb0db5 on requests:master.

[ryanhiebert]

Thank you for the nudge. This latest commit fixed the bugs, but I still haven't added any regression testing.

[ryanhiebert]

I am working on the regression test, but looking at @Lukasa 's comment to me about checking on auth that we do, I'm not sure how to do any of that. My impression is that there's nothing really to test with that, because all of the current tests should handle those cases.

The only thing I'm seeing being worth adding AFA tests is the regression test to make sure that the .netrc isn't used, even though thats the default for requests.

[ryanhiebert]

Digging into the session code of requests a bit more, it looks like there's a trust_env setting that defaults to True.

http://docs.python-requests.org/en/master/api/#requests.Session.trust_env

Would settings trust_env to False be a better option than the noop auth? It'd certainly work for my use-case. The downside is that it looks like it also bypasses any proxies, so it's not strictly relegated to just avoiding using the .netrc files, though I'm not sure how much of an issue that caveat is.

[ryanhiebert]

OK, I've added a regression test suite, that inherits from the main test suite but sets up a problematic .netrc file before it begins, and caused 2 of the tests to fail (so they are good regression tests). I've also added an entry in the HISTORY file.

I've kept the no-op auth as the approach, because I'm a bit hesitant to incur the side-effect of bypassing proxies, and because the no-op lambda is already short enough to make me happy.

[singingwolfboy]

Looks great to me! @Lukasa, what do you think?

[ryanhiebert]

What are we thinking about this?

[singingwolfboy]

@ryanhiebert if you rebase to resolve the conflicts, and the tests still pass, then I'll merge this.

[ryanhiebert]

Thanks, @singingwolfboy . I've done that.