Mauricio Scheffer

I think this is caused by https://github.com/mautic/docker-mautic/issues/90 (and there's quite a bit of discussion about this on that issue)

Thanks @hkaj , that is the workaround I'm using. But as I said, that only tells me that "there's something wrong" and takes away all the monitors and metrics I've...

Looks like you can do bind before connect with `requests` after all. I asked on twitter and got this gist from @agramajo https://gist.github.com/agramajo/6145047aa49af419c8d2600797cfe752

I recently got [micromamba merged in nixpkgs](https://github.com/NixOS/nixpkgs/pull/129583), which is somewhat simpler to operate than conda, maybe worth trying that instead?

Just found https://github.com/Netflix/metaflow/blob/master/metaflow/metaflow_config.py so adding a few more comments in there and linking it from the documentation would be great too.

One solution would be changing the `option` representation: instead of using `null` for None, which causes the ambiguity, use `"None"`. And probably `Some x` should be also represented in JSON...

Actually there's no need to wrap the `string option` in a separate type to reproduce this. Simply adding `yield testProperty "string option" (roundtrip)` to https://github.com/mausch/Fleece/blob/master/Tests/Tests.fs#L207 fails.

The last case you mention is slightly different: when using `.@?`, `None` corresponds to a missing field and `Some x` corresponds to a present field with value `x`. It can't...

BTW you might want to use `Success` / `Failure` instead of `Choice1Of2` / `Choice2Of2`: easier to type and easier to tell from each other when reading it.

Yes, I think that makes sense. It's either that or throwing an exception. Implemented that change in b8964b019f185c8ed3932f9c9dac837a07522423 . As for optional serialization elements, personally I'd rather be explicit about...