Kai

@sts In the example you provided, where does IP address come from? Is it just a field in the documents being aggregated? You might be able to do this with...

Oh, nice find! We use enrichers purely on the frontend, so we haven't hit this problem. I'm wary of flattening the data though - maybe it'd be better to replace...

Thanks for adding hints and cleaning up all the docstrings. * I can resolve any conflicts, so no need to worry about that * Could you elaborate on this? Are...

Got it. That sounds fine then. For `re`, that block of code is indeed for Py2 support - `Union[re.Pattern, str]` should do the trick with its removal.

Yup, the repo uses Google-style docstrings (https://sphinxcontrib-napoleon.readthedocs.io/en/latest/example_google.html). From a quick look at that page, it seems that Sphinx supports/knows about native type hints now. As such, I think we could...

Thanks for your work! I've merged the changes into `deprecate_py2`. I'll do a review of the library and then merge into main.

Bump (although it looks like there's work going on in the linked issues)

Relevant thread: https://r2c-community.slack.com/archives/C018NJRRCJ0/p1630349798014700

For the `No SLF4J providers were found.` error, should a default logging provider (`slf4j-simple`, perhaps) be included with the distribution?

Bump! The manual token exchange appears to be more reliable at the moment. Could support be added for using this flow? I can make updates to the PR as appropiate.