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
To prevent the issue in #1247 convert some cs2_area_fraction_top to nans) from happening again(at least prevent it from happening a lot), we should check the impact of one PR on the fields.
In details, we should check
The change in the number of nans.
The deleted or added fields.
These can be done by writing scripts of actions.
The text was updated successfully, but these errors were encountered:
Assuming that we have some functions to iterate through the dependency tree and order the plugins by their "genetic relationship", then we can search for the oldest change of lineage, and compare the data produced.
Here is the plan:
search for deleted plugins, and report their names
search for the added plugins and report:
+. the fraction of nan(0 and -1) of float(int) fields
search for the oldest changed(in lineage, because the __version__ will be tracked by lineage) plugin and report
+. the fraction of nan(0 and -1) of float(int) fields
+. the mean absolute change in each field
+. whether the oldest lineage-changed plugin is really the oldest results-changed plugin by comparing the result of the second oldest layer of plugins
if possible, compare the change codes, hint at which plugin is really changed by programmer
Thanks Dacheng, is there a database somewhere which tracks the lineage hash of the previous version of plugins, so that we can compare it to the lineage hash of the plugins of a PR?
To prevent the issue in #1247 convert some
cs2_area_fraction_top
to nans) from happening again(at least prevent it from happening a lot), we should check the impact of one PR on the fields.In details, we should check
These can be done by writing scripts of actions.
The text was updated successfully, but these errors were encountered: