Mutliple Jupyterlab sessions sharing state?

[Brendan Barnwell]

I have multiple conda environments with different setups.  Sometimes I want to have Jupyterlab sessions from different environments operating on code in the same directory (e.g., to test how the code works with different versions of packages).

Here's what happens: I open a Jupyterlab session in a particular working directory.  I open some notebooks.  I then shut down the session.  I then open a new Jupyterlab session from a different conda environment, but within the same working directory.  I expect this session to be totally independent of the first, but it seems not to be.  It starts up with the same files open in tabs as I had in the previous session.

Is Jupyterlab somehow storing state in a manner that does not take account of where it was run from?  If so, is there a way to get it not to do that?  I'd like to be able to have two totally independent Jupyterlab sessions, both with the same working directory, but not affecting each other in terms of this UI stuff like remembering which documents were open.