Casper da Costa-Luis

@0x2b3bfa0 we have a (doubly hidden) `publish --url` now, but we're missing a public API. i.e. we still need `cml comment {create,update} --publish-url`

The problem with whitelist-only is there are a few other things that a user would likely want preserved too... - `NVIDIA_VISIBLE_DEVICES` - `CUDA_VISIBLE_DEVICES` - `LANG` I'd prefer blacklisting CML config...

I have this issue on on-prem as well as cloud...

instance termination on error seems sensible (to avoid spiralling costs). Not sure about debugging though (some users may prefer to not have runners shut down? maybe add a flag to...

While reviewing the 72h -> 35d GHA self-hosted runner timeout change (#1064), stumbled on: > A self-hosted runner is automatically removed from GitHub if it has not connected to GitHub...

How about forms? Would be an even better upgrade. https://docs.github.com/en/communities/using-templates-to-encourage-useful-issues-and-pull-requests/configuring-issue-templates-for-your-repository#creating-issue-forms However my tuppence is such automation may not be that necessary given how responsive we are to new issues. Automating...

> you could invoke it twice using labels for your two jobs you want to run in parallel you could also invoke it N times with the same labels but...

> shut down the runner every time we do experiments and run it again once we're finished. Sounds like the best solution in your use case (on-prem mixed-use, i.e. local...

I think either: - fail nosily on conflicts, or - avoid conflicts à la [SO/70205896#3896283](https://stackoverflow.com/a/70205896/3896283)

> Use the [image description](https://spec.commonmark.org/0.30/#image-description) to store arbitrary data encoded as Base64 :nauseated_face: