Add defaults

[flying-sheep]

Supersedes #103, fixes #48

In the previous PR, I added them to types, this time I’m adding it to the string. I think both have arguments pro and contra. Should it be configurable?

[gaborbernat]

I'd like to have it by default off. Otherwise, I'm happy to also allow it to be configurable where it's added.

[flying-sheep]

OK, I need help, I have no idea what’s going on with those "s and *s.

Why does only my test case fail on Python 3.9 even though I copied the setup from test_sphinx_output_future_annotations?

[gaborbernat]

OK, I need help, I have no idea what’s going on with those "s and *s.

Why does only my test case fail on Python 3.9 even though I copied the setup from test_sphinx_output_future_annotations?

That's half the work in this project, to figure out the differences between Python versions and make it work everywhere. I'll have a look at some point, but no idea when I'll have time for this.

[flying-sheep]

Ugh, I figured it out. This will bite you in the future, I hope you find a way to make it less flaky. I filed #194 to explore this.

This PR is now ready! Changelog, readme, code, tests!

[gaborbernat]

Ugh, I figured it out. This will bite you in the future, I hope you find a way to make it less flaky.

I filed #194 to explore this. But this PR is now ready!

Sorry, I'm not open to a fallback issue and working around the test. We should figure out here and now why this is the case and avoid it.

[flying-sheep]

I’m avoiding it by not using that syntax, in the same way your main test does. This PR is completely independent of this problem now, I just happened to run into it while developing this PR.

I’m happy to help, but please don‘t drag this unrelated PR into it.