Fixes the two problems that appeared in #34 (closed):
conda activate baseonly works for interactive shell sessions (see here for an explanation). By having set
#!/bin/bashto define the execution environment the activation fails... and hence JupyterLab does not start. I have implemented a proposed workaround in both the HLRN and Nesh job scripts.
Also, the script fails to retrieve the remote JupyterLab URL if a user has configured another security mechanism (those that cause the token to disappear in the URL, e.g. a password). I have implemented a more general way that doesn't look for the token, but the remote hostname in the URL now. This should be more robust. Currently, this is only implemented for the Nesh job script, though.
Otherwise, I have added/updated descriptions, removed the explicit specification of the number of nodes for the Nesh jobscript (this is set to one per default by the job scheduler and it doesn't make sense to have a larger number) and set the JupyterLab base directory to
$HOME as in the HLRN job script.