There should be issues

[mpnally]

Every healthy open source project should have issues, otherwise it is being ignored

[handrews]

@mpnally looks like commit activity here has tailed off somewhat. Rapier has a lot of really interesting concepts that I've not seen implemented elsewhere. I've been looking at writing a system with some similar ideas based on JSON Hyper-Schema, but could potentially be interested in using Rapier instead if it's being actively maintained and expanded. Or would be if someone were filing issues :-)

What's the current status of the project?

[mpnally]

Thanks, Henry. Rapier is a sort of hobby project. I still use it personally to design APIs. I have done some maintenance on it, but it could use work. If you showed interest, it might reinvigorate me. I don't know much about JSON hyper-schema, but when I looked at it, it seemed to be trying to infer links that are not in the representations from other data that is. That wasn't a problem I was interested in — all my APIs already have the links explicitly in the JSON representations. My interest was in helping design more APIs like this. Maybe I have not understood JSON hyper-schema. What problems are you interested in?

[handrews]

Hi @mpnally thanks for the quick reply! Sorry I've been a bit distracted this past week.

Hyper-Schema does allow for building links from data in the representation, although it can work with links that are already present by using a trivial URI template of "{linkPropertyName}" which just replaces the entire URI template with the contents of the linkPropertyName property, which presumably holds a URI.

The main part of Rapier that interests me is that it builds on the assumption that APIs use HTTP correctly, rather than trying to describe things that HTTP should clearly imply/constrain already. I think Rapier and Hyper-Schema are taking philosophically similar approaches here.

Secondarily, the idea of leveraging OpenAPI/Swagger for documenting an API approach that many find familiar while introducing a more innovative approach to hypermedia is a great one. I'd been thinking some about how Hyper-Schema and OpenAPI could fit together, but had not developed the idea much.

I'm also particularly interested in building generic clients. However, if I understand the Rapier generated clients correctly, they are still specific to particular APIs, even if they are generated in a generic manner. So that part may not line up.

I need to read a little more of what you've done here- definitely some great ideas and at minimum some inspiration. I think it probably makes more sense for me to see if I can get a bit further with Hyper-Schema as there have been more people showing up to that. That's not a technical judgement of Rapier, just me trying to figure out which system is most likely to produce an active ecosystem. Hyper-Schema isn't there yet, and is behind Rapier in terms of implementation AFAIK, but there might be more people's work to leverage in the near future.

(BTW you are welcome to delete these comments if you don't want my hyper-schema musings cluttering up your issue discussion).