mirekphd

More importantly, critical vulnerabilities are also detected in `the vm2` node module used in `v4.16.1`: ``` vm2 3.9.19 GHSA-cchq-frgv-rjh5 Critical /usr/lib/code-server/node_modules/vm2/package.json vm2 3.9.19 GHSA-g644-9gfx-q4q4 Critical /usr/lib/code-server/node_modules/vm2/package.json ``` Due to them...

> --config Yes, so say [your docs](https://coder.com/docs/code-server/latest/FAQ#how-can-i-reuse-my-vs-code-configuration) (at least for `--user-data-dir`), but neither of these two options seems to be available in `code-server` binary: ``` (base) coder@5800d4264eeb:/home/jovyan$ code-server --version 1.79.2...

> neither of these two options seems to be available in `code-server` binary: Turns out there are *two* instances of `code-server` available at runtime, and the one which is executed...

@knqyf263 any reproducible example would be appreciated enabling us to scan an image (preferably an unpushed one or one from a local insecure registry at the very least) under a...

The patch from #141 seems to fail to work - empty `/nbextensions` can appear just as they did previously. I tried to install the configurator from master branch at Github...

Related SO issue: https://stackoverflow.com/questions/68208833/jupyter-lab-does-not-load-jupyter-nbextensions-configurator-but-jupyter-noteboo

> trivy will prefix `mookas/junk` with `docker.io/`, resulting in `docker.io/mookas/junk`. It won't find this image locally because you have it stored locally as `mookas/junk` If that were the case, then...

@cboettig This works, but only in docker. It would be a security hole if one could set an arbitrary user ID by means of an environmental variable, which is why...

And the /etc/password editing script added to entry point (where the final UID is known) comes verbatim straight from the OpenShift documentation: https://docs.openshift.com/container-platform/3.10/creating_images/guidelines.html#openshift-specific-guidelines (see Support Arbitrary User IDs). Make sure...

> Training time 7.5h on 3090. I suppose Kaggle grandmaster's lesson of using 5-fold cross-validation also for DL models is out of the window then:)