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
Pymatviz changes frequently - this is not a problem, apart from the fact that it pymatviz is not pinned in matbench discovery. This makes it tricky to keep old installations working, because released versions of matbench e.g v1.3.1 do not specify an upper bound for pymatviz. For example, this commit: janosh/pymatviz@ab691bb breaks all installed versions of matbench_discovery which are not main, because the package is not pinned.
I think given pymatviz is not a library which is super general (like torch or numpy), it might be safe and not disruptive to end users of matbench to pin the version. This is related to #145
Thanks!
The text was updated successfully, but these errors were encountered:
that's a good point! admittedly it's been the wild west of breaking changes, especially around end of last year. i think they'll be less frequent going forward, at least on the pymatviz side. you're right though, a strict version pin might help as each commit in this repo should be compatible with the latest pymatviz at the time. i'll add that
Pymatviz changes frequently - this is not a problem, apart from the fact that it pymatviz is not pinned in matbench discovery. This makes it tricky to keep old installations working, because released versions of matbench e.g v1.3.1 do not specify an upper bound for pymatviz. For example, this commit: janosh/pymatviz@ab691bb breaks all installed versions of matbench_discovery which are not
main
, because the package is not pinned.I think given pymatviz is not a library which is super general (like torch or numpy), it might be safe and not disruptive to end users of matbench to pin the version. This is related to #145
Thanks!
The text was updated successfully, but these errors were encountered: