Jeremy Tuloup

JupyterLab has an internal function to do this: https://github.com/jupyterlab/jupyterlab/blob/c6744e548c7b2790e4c6a4d41a60aae03f22c62a/jupyterlab/federated_labextensions.py#L361-L398 ```py def _get_labextension_dir(user=False, sys_prefix=False, prefix=None, labextensions_dir=None): """Return the labextension directory specified Parameters ---------- user : bool [default: False] Get the user's...

Thanks @michaelweinold for reporting. https://github.com/jupyterlite/jupyterlite/pull/1024 added support for chunked uploads, but maybe it does not support gzip files: https://github.com/jupyterlite/jupyterlite/issues/534 (would need to investigate)

> Reading the gzipped files (for instance, once they have been created with [the test function](https://github.com/jupyterlite/jupyterlite/issues/1153#issue-1899772800)) does not work. Only when they are programmatically created, or also when they are...

It could be yes, would need to investigate a bit further to confirm.

Thanks @michaelweinold for checking. I haven't investigated it yet, but will hopefully have a look soon.

It's probably still related to the custom contents manager used in JupyterLite, which hasn't changed much for the past release. So still something to be investigated and to be released...

Thanks @martinRenou for opening the opening, this looks like a good summary of the current service worker issues :+1: > At some point, it would be nice to ditch the...

> I am suggesting putting the service worker registration behind a flag So that means users will not have their contents available from the kernel by default anymore? How about...

> I think that if we: > > * document enabling shared array buffers for having file-system access support > > * document enabling service worker for PWA > >...

Which is why putting caching behind an opt-in flag could be a good first step, and could be shipped in a `0.2.x` release quite quickly. That way we could also...