Maven Central

Pros

• End users won't need to add an extra repository to their build files
• The namespace/package of core parsers will be consistent with jtreesitter

Cons

• Parser authors will need a Sonatype account and GPG key
• Requires a new reusable workflow

JitPack

Pros

• Parsers authors don't need to do anything special
• Doesn't require a new workflow

Cons

• Requires an additional generated file
• End users will need to add an extra repository to their build files
• The namespace/package of parsers will be inconsistent with jtreesitter

I'd like some user opinions regarding the packaging.

/cc @stackoverflow @bioball @Emmeral @maloewe-ona

So if I understand this correctly, the actual binaries will not be uploaded to the remote repository but only the bindings class and the user of the bindings needs to supply the binaries themself?

I'd like some user opinions regarding the packaging.

I don't have a specific opinion on the Maven Central vs JitPack topic. For my use case, I would need to modify the pom in both cases.

The cons of Maven Central seem significantly worse to me than those for JitPack.

Perhaps, but something like this is not possible in JitPack:

So if I understand this correctly, the actual binaries will not be uploaded to the remote repository but only the bindings class and the user of the bindings needs to supply the binaries themself?

I'm leaning towards Maven Central after all since it allows us to fully control the release process.

Just to clarify:

Parser authors will need a Sonatype account and GPG key

Does this mean that individual grammar / parser authors need to setup and maintain these bits over the long term then?

They also need (if they want) to set up and maintain tokens for npmjs, pypi, and crates. The only extra here is the GPG key.

I wonder about the practicality of this kind of arrangement over the longer term. It seems to me that many a grammar / parser author has lost interest after a certain point. In a nicer world, may be there would be two or more folks looking after a given grammar / parser for robustness (though OFC that's a larger attack surface as well...).

I've been ok with looking after a few grammars, but I have no interest in setting up and properly looking after additional accounts and/or tokens (and thereby becoming a slightly more juicier target). Perhaps this is a minority position though.

Although, now that I think about it, those can also be installed directly from git.
That is sort of possible with JitPack but not (easily) with Maven Central.