Hey, we've been using pants v2 for about two weeks now for all our ML/DL stuff, and it's been awsome 🔥🔥, no major problems identified, just by following the docs, I was able to have a working linter/formatter and ci pipeline easily so kudos for that too, one thing that I miss dearly for a data science project is to use jupyter notebooks, I know that Ipython is supported so jupyter should theoritically should easy to add too ( maybe as a plugin ? ). Another minor thing is the time taken to re-resolving of all dependencies when running tests or binaries, (usually 10 mins per target, with multiple targets having the same deps...), but as I've it's only minor, since having a warm cache usually speeds up stuff greatly

Yay! Great to hear about your experience! Agreed on Jupyter. Did you have a v1 plugin to do this? Also agreed on the time to resolve dependencies. Are you using 1.29? We made an initial improvement to speed things up through better caching. But it’s still not ideal. This is our biggest performance priority before we launch Pants 2.0

nope, no plugin 😅, I used to do a hack by just adding the pyprep deps and sources to sys.path on top of my notebook .... as I've said just a hack ahah Cool for the speedups, looking forward to it, yep we are using rc0 of 1.29 right now

Thanks for the feedback @rapid-crayon-8232! Resolver performance is indeed our biggest priority. Jupyter support has come up before, I'd be interested in hearing what you think this should look like, from a user perspective.

the notebook thing is kinda complicated...they main difference between an ipython repl and a notebook is that the notebook is actually multiple python processes: one server/launcher that will start "kernels" (other python interpreters) on demand. It's fairly easy to build an juputer notebook server executable with the

jupyter.notebookapp:main

entry point, but this server can't successfully start other python kernels

this is definitely an important area. i’d love to help get native support in place.

I’d advocate it being something like a 2.1 feature so that we can focus on releasing 2.0 sooner and ensuring it’s stable and polished

this is an important feature, but it's covering a totally new use case, i wouldn't make it into a blocker for something