Jimmy Jia

@vphilippon Would you consider a version of #169 updated again for the current master? Or at least the part the resolves the commit for editable VCS URLs?

This seems like it has some risk of unpredictable behavior in the case where multiple schema classes define options classes, though. It seems like you can still end up with...

I'm thinking something like: ```python class Opts1(SchemaOpts): pass class Opts2(Opts1): pass class Schema1(Schema): OPTIONS_CLASS = Opts1 class Schema2(Schema1): OPTIONS_CLASS = Opts2 class Schema3(Schema1): pass class Schema4(Schema3, Schema2, Schema1): pass ```...

Also, if the ARN does not exist, it appears that we actually get an `InvalidParameter` error from AWS with message: `Invalid parameter: TargetArn Reason: No endpoint found for the target...

That's not quite correct. If `CustomUserData` is set, then AWS will reject the request.

It depends how closely you want to follow semver. Probably it would be more confusing than it'd be worth for the version of babel-standalone tracking Babel 6.13.x to be 7.0.0....

Probably the only thing we can do here is to deprecate those, if we want to keep semver.

I'm not sure how I feel about moving the converter off of "meta". It seems to me that it intuitively should be "next to" the declaration of the model or...

Is the only thing this saves in practice not having to use a custom schema opts subclass? I guess I don't see offhand why it should be different from, say,...

Well, would it be reasonable to do both? Can the converter look at the schema to get the type mapping? It feels more weird for type mapping to work differently...