update hyperlinkrelated to gracefully support nullable / non-required sqla relations

[feigner]

I have nullable relations on my SQLAlchemy models that blow up when deserializing HyperlinkRelated. Patch allows for graceful failure when base field is not-required.

Similar to the issue listed here: #18

[antgel]

@feigner Lots of interest in this according to various Issues (e.g. #18 and #68). Perhaps you can rebase against latest master, then @sloria would you be able to merge it?

[sloria]

Sure. @feigner Would you mind getting this up to date with dev (rebase or merge are fine)?

[feigner]

Yes, I can take care of this soon Sent from a TI-89.

…

On Feb 11, 2019, at 08:05, Steven Loria ***@***.***> wrote: Sure. @feigner Would you mind getting this up to date with dev (rebase or merge are fine)? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[feigner]

Allllrighty @sloria & @antgel, just rebased against dev and updated accordingly. Looks like tests are all now passing 👍

[sloria]

@feigner Thanks for getting this up to date. I'm unsure how this will affect partial loading. Can you add a test for it?

[feigner]

Happy to write some more tests. I've read the linked docs, but I'm not entirely sure what you'd like me to verify. Can you give me an example use case of what you'd like to see?

[sloria]

Oh, didn't realize this only changes serialization behavior, so partial loading isn't relevant here.

However, I'm not sure about the if not self.required check. marshmallow does not validate on serialization (by design), so _serialize shouldn't do any "required" validation.

I think the most consistent behavior would be to serialize None to None, as the other fields do.

[sloria]

Upon looking this over again, I think this isn't the right approach for the reason I mentioned above. required and missing should only be used in deserialization. The proper fix is actually simpler--just serialize None to None. I've done this in #125.