Jupyter vs Zipline container

The docs suggest using the jupyter container when editing code via VS code. I've found this to be a less ideal experience than connecting to zipline, especially when using custom packages to support common functionality between algos.

I think a somewhat recent version removed the special environment for zipline. Unfortunately, this has the side-effect of making co-development of packages and an algo that references them basically infeasible in jupyter since the zipline environment is complete isolated. Furthermore, basic quality of life things like autocomplete are missing in jupyter for custom packages.

Am I missing something obvious about how to use jupyter, or is it expected that algo and package development in zipline should still be the suggested route?

This question is a little broad and hard to answer but here are a few points:

  • Containers are isolated from each other, so if you want packages to be accessible in multiple containers, you must install them in each container. This has always been the case.
  • Since version 2.6.0, the same basic environment is installed on each container, so it doesn’t really matter which one you attach to. jupyter is a reasonable default option but it’s fine to use zipline.
  • There no good reason why autocomplete should work in the zipline container and not in the jupyter container. I would check that you have the necessary VS Code extensions installed in the container and have installed the package in each container where you want to use it.