Open aidanheerdegen opened 2 months ago
One difficulty is how we save this information.
It would be good to be git commit
this log to the control experiment git
repo. That is tricky: do we put up with random commits? The path to the committed file might reference an archive
path that doesn't exist in the future when files in scratch
are purged.
With the new
sync
features it would be good to log what files were sync'ed and where. This is a good idea so users can come back to an experiment and discover what was done, but also from the point of view of keeping track of output and restart files.The location of restarts is important as this makes it possible to add this information to a database to make it easier for users to create experiments that are based on previous experiments.