Matthew Kenigsberg
Matthew Kenigsberg
> Also should we prefer newer instances of the same version of a package in general? I think that is not taken into account at all yet I believe we...
> Personally I also don't think that getting a newer package that is broken if there are older version that are not broken is that bad (especially since older instances...
What's the current flow for getting a token? Do we need to add a command for that?
Args the watchdog needs: - disable metrics: will be needed by so might as well add an env var - PID: accessible in Bash (`$$`) - Log dir: already passed...
Beware of merge conflicts with https://github.com/flox/flox/issues/2134
> > launch it in the activation script after sourcing hook.on-activate > > Can you describe why it needs to come after `hook.on-activate`? I don't think the `flox-watchdog` depends on...
I think the idea was that ultimately we may start `process-compose` as part of `start` and then start the watchdog immediately after that. But for cleaning up activations.json it would...
> I didn't have this in mind. Just to make sure that we're on the same page though; I had assumed that the watchdog would initially be parented to the...
Chatted with @limeytexan sync, the comment he left earlier was a typo Summary - We should launch the watchdog after running hook.on-activate, because we may want the watchdog to run...
Ended up starting the watchdog in Bash for reasons explained in https://github.com/flox/flox/pull/2276/commits/a916936f3adfea22eed3e4eff4ff6eae150a259b