Arne Brasseur
Arne Brasseur
I may be misunderstanding something. Isn't it enrich-classpath that adds `src.zip`? because that definitely ends up on the root classloader's path.
Thanks for clarifying @vemv, it seems I did indeed misunderstand enrich-classpath. > one thing is performing classloader tricks or breaking clojure.java.classpath/classpath; another is placing items in the classpath beforehand via...
I did some more testing on this, and my main impression is that we should start getting rid of some of the noise. `clojure.main/repl` adds a new `DynamicClassLoader` to the...
Interesting observation today. `future`s are scheduled on the `clojure.lang.Agent/soloExecutor`, which is a fixed size threadpool. So calling `future` may create a new thread or reuse an existing thread. Threads inherit...
With the above changes to cider/cider-nrepl/nrepl I can load the middleware, and invoke a `cider-version`. Not everything is working yet (still trying things out and investigating) but this is a...
Question regarding the sideloader design for @shen-tian. It seems like there's no way to distinguish between "file not found" and "empty file"? Maybe not the most common edge case but...
With the last two PRs this actually seems to pretty much work as intended. It's a little slow (so might be worth considering caching responses, since currently every clj file...
in any case I think it should be opt-in, the current jack-in approach works fine and has benefits over the sideloading (speed, having resources actually on the classpath).
Here are some quick PoC experiments: https://gist.github.com/720798947e999a2142390a702a9ebe0b - shelling out to a suitable tool for fetching the jar and returning a classpath We have a lot of options here these...
I'm continuing to dogfood this, and have some more changes to nrepl incoming, including the caching we talked about. What I'm seeing in real-world usage though is that this slows...