Pass CLASSPATH variable through to scala interpreter

[tonycpsu]

The scala REPL can pull in JAR files from the CLASSPATH like so:

$ CLASSPATH=/foo/bar/baz.jar scala

However, trying the same with jupyter-console doesn't seem to work:

$  CLASSPATH=/foo/bar/baz.jar jupyter-console --kernel=scala211

I know about load.jar("/foo/bar/baz.jar"), but it's much easier to use the environment variable than to have to load each jar individually.

[alexarchambault]

It can't be done as is, the kernel somehow filters the classpath it is started with, in order not to expose all of its own dependencies. Having some kind of predef, like Ammonite does, would allow to solve the problem by putting load.jar lines in it.

[maclockard]

Is it possible to just use the Ammonite predef, or will there need to be a separate predef created to work with the kernel?

[espears4sq]

I have the same question as @maclockard . How do we define predef scripts that will be picked up when running both of jupyter notebook (and selecting the scala kernel), and jupyter console --kernel scala?

I tried modifying the ammonite predef in ~/.ammonite, but neither method of launching jupyter-scala will load or execute any files from that path at all. Is there a different location for the jupyter-based equivalent predef?

[alexarchambault]

Closing this now that the upcoming version is about to become the default one. It accepts a --predef option to run some code upfront.