Initial outline of reference implementation.

[jeromekelleher]

WIP putting together a reference implementation in Python. The terms-of-reference for this are still coming together in my head as it develops, but the current idea is to make it a strict but unhelpful parser (i.e., don't expect nice error messages).

I'm avoiding using JSON schema as I want to implement it as someone reading the spec without having a JSON schema implementation to lean on.

There's a question about how much CI etc we do with this. My vote would be to keep it fairly minimal.

[grahamgower]

Looks good so far!

There's a question about how much CI etc we do with this. My vote would be to keep it fairly minimal.

I think it will be useful for the demes-spec repository to include a set of test yaml (or json) files. There would be two sets, one for valid input , and one invalid input. The intention is for these to be available for the development/testing of new implementations. This reference implementation could then be tested against those files.

[jeromekelleher]

It would be good to get some input on this here @grahamgower, @apragsdale, @molpopgen. There's still a fair bit to be done, but I think the general outline of how we do the "hard" bits is reasonably clear. It took a few attempts to boil it down to this.

I'm hoping the code is clear, so I won't explain it. Start at the Graph.parse method - any input much appreciated!

[apragsdale]

Looks really great so far! This was really useful to read through, and will certainly help also with cleaning up the demes-python implementation I think.

I suppose the next tricky part will be figuring out everything that belongs in graph.validate(). I really like how you've put together the default attribute inheritance here.

[grahamgower]

Looks very clean, definitely we should port stuff into demes-python where it's easy to do so.

[molpopgen]

I'm still going over the details, but this looks great!

[jeromekelleher]

I think this is ready for final review and merging now. There's a few loose ends to tidy up and we need to add in a lot more examples (see #27), but overall I think it's fairly solid. I suggest we merge this fairly soon, and then follow up with the corresponding updates to the JSON schema, spec wording, etc. We should also follow up to add basic CI too (basically, run tests, check that coverage is 100% [no need for codecov, etc, hopefully], and run mypy).

[jeromekelleher]

Update here: I've added a bunch of example files like examples/x.yml and examples/x.resolved.json. These are supposed to be examples of an input yaml file and the output fully qualified JSON, which should be a good resource for parser writers. It'd be nice to build this out into a sort of "compliance suite" by adding more examples, although I guess we'd need to add a bunch of files that provoke errors too to do it properly.

Anyway, don't bother going through the .resolved.json files too much, as they are auto generated.

[grahamgower]

Looks great @jeromekelleher! Just some minor comments below.

[jeromekelleher]

Great, thanks @grahamgower! I've tacked on a few commits to deal with these and opened issues for the stuff that's a bit more work. I'll squash these before final merge.

@molpopgen, @apragsdale, are you happy to merge this?

[molpopgen]

Happy to merge

[jeromekelleher]

@apragsdale?

[apragsdale]

Yes, happy to merge - nothing jumps out to me that I could see!

[jeromekelleher]

Great! All squashed and ready to go - I'll let you push the merge button @grahamgower.

[jeromekelleher]

(I merged this as I want to get on with #40)