bufdev

Yea makes sense re: lazy handling. This is definitely a scenario I'm interested in seeing handled, let me know if there's anything I can do to help (although assuming you...

The source I have is a FileDescriptorSet, sometimes off the wire, sometimes created with third party lib (jhump/protoreflect) - for JSON marshalling though,I end up calling protodesc.NewFiles so I can...

I could put together an example of what I’m doing in code if it’d help, ie extracted from the complication of bufbuild/buf (which is only the public half anyways)

Oh to be clear - it didn’t work :-) solving this/looking into it in the detail of this issue was low on the priority list, as for bufbuild/buf’s current public...

Ah yes - confused myself on reading your response - yes that’s actually a trivial amount of time in comparison, but...don’t think that solves the protoreflect issue right? Although as...

This sounds like something that would be great in documentation, and probably should be added there. Then, as users will be informed of the choice, let users decide what they...

I empathize with what you're trying to do here, and I know it has sound basis and great intentions - users should not be relying on the stability of JSON...

> We're never removing the package. Yes, but users are being asked to upgrade to the new v2 implicitly regardless, as of 1.4.0 - you could use the old `jsonpb.Marshal`...

> The right course of action is to focus energy on getting a canonical format defined and implemented. This would be a fantastic outcome - is there anything we can...

Going back many years in my CS life, heh. Some potential leads: - https://tools.ietf.org/id/draft-rundgren-json-canonicalization-scheme-05.html#rfc.section.3.2.2.3 - https://www.ecma-international.org/ecma-262/6.0/index.html#sec-tostring-applied-to-the-number-type (linked from 3.2.2.3 above)