Mingun

I would like to: 1. have tests for that, at least this one in `serde-issues.rs`: ```rust /// Regression test for https://github.com/tafia/quick-xml/issues/683. #[test] fn issue683() { #[derive(Deserialize, Debug, PartialEq)] enum ScheduleLocation...

Thanks @leftmostcat and @Xiphoseer !

Currently this is not implemented and it is already tracked in #313. Contributions are welcomed!

That's generally impossible without changes in serde, because you cannot get any meta-information from the deserializer generically. Actually, I tried to implement this but that required changes in serde and...

Yes, that is expected. This is the same behavior as in `serde_json` (yes, it is asymmetric de vs. ser). If you want to skip field serialization, add `#[serde(skip_serializing_if = "Option::is_none")]`....

Main question now -- do you agree that DTD parsing can be moved to it's own crate, or you prefer to keep it as a module inside quick-xml? If use...

> You said you actually did notice a compile time difference - how much? I was not clear enough. The point is that if create a new module inside quick-xml...

This is because we do not normalize attribute values according to the [specification](https://www.w3.org/TR/xml11/#AVNormalize) yet, where all whitespace characters would be turned into space, but deserializer already written with that in...

No, this is more related to #371. Normalization of attributes does not handled correctly yet neither in `Writer` or serde. The only workaround is to use `#[serde(serialize_with)]` and normalize value...

Uh... doesn't think about that. One possible workaround for that is to make two-step escaping: - using `serialize_with` replace necessary symbols to some markers (for example, `\n` -> `\u0001#xA;` -...