You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have greatly appreciated the new features for restoring runs (#245) which have been a big timesaver for HPC tangos applications.
They however interact unexpectedly with the --force keyword. Even if this keyword is passed, if the Tangos runs has previously crashed/was terminated, the calculation resumes from its previous state rather than recomputing from start as expected from setting --force. You can of course delete the cache folder to avoid this, but the point of --force in my mind is to supersede time-gaining manoeuvres (e.g. don't recalculate if properties already exist).
Was this a wanted behaviour from the resume feature, or an oversight?
Martin
The text was updated successfully, but these errors were encountered:
The flag you are looking for is actually --no-resume. I kept this separate from --force since they seem logically distinct to me, and one might want one without the other. That said, I think the advice to delete all the resume info is probably not the best — the message should probably point to the flag rather than the brute-force approach.
Hi,
I have greatly appreciated the new features for restoring runs (#245) which have been a big timesaver for HPC tangos applications.
They however interact unexpectedly with the
--force
keyword. Even if this keyword is passed, if the Tangos runs has previously crashed/was terminated, the calculation resumes from its previous state rather than recomputing from start as expected from setting--force
. You can of course delete the cache folder to avoid this, but the point of--force
in my mind is to supersede time-gaining manoeuvres (e.g. don't recalculate if properties already exist).Was this a wanted behaviour from the resume feature, or an oversight?
Martin
The text was updated successfully, but these errors were encountered: