bowbahdoe
commented
2 years ago Hey @alexarchambault!
So my goals with this project are
- to make a dependency resolver that could eventually be upstreamed into openjdk
- pave the way for a build tool to be built in to openjdk for Java / easier experimentation toward that effect.
So coursier/interface is part of the way there in that it
- Has no runtime dependencies
- Has a more sensible API than Aether
- Handles concurrency better than Aether (personally I ran into this a bit more than I wanted to and its not the clojure folks fault)
But it still doesn't fit the bill in that
- Doesn't have a
module-info.java(maybe that would be easy to add - idk) - There is still a "supply-chain" dependency on Scala and the Scala ecosystem
And if we are doing it with that aim and without the burden of backward compatibility, there are parts of coursier (I suspect) can be simplified in translation
- Monad based resolution might not be necessary with virtual threads
- Public API surface might be able to be trimmed down some
- Ivy support might not be important enough to carry over without the context of SBT
- The Java 11 HttpClient might obviate the need for a lot of the custom
URLConnectionandAuthenticationlogic.
The other folks in this repo are students from RIT who are working on this with me as part of their senior design project. We're at a point where the reverse engineering has gotten tricky, especially with me being the only one with Scala exposure.
We all have a bunch of questions, if you have the patience.
npetrangelo
I stumbled upon this repository via https://github.com/coursier/coursier/issues/2549.
Not sure what's its overall goal, but FYI, if that can help, coursier already has a pure Java API here: https://github.com/coursier/interface. coursier/interface uses shading and proguard to create a pure Java JAR, exposing a subset of the coursier API.