(Based on discussions on @xgui3783 and I, @AhmetNSimsek. @xgui3783 please feel free to edit if I missed sth, or comment new points.)
Version deprecation could involve a few possible reasons:
Previously undiscovered bug fix
Significant changes in the preconfigurations
This entails a possible case of data in neuroglancer.humanbrainproject.eu dies or it needs to be migrated for another reason.
Or KG availability
I approached @xgui3783 with the following initial idea:
For a simple check of the newest version, I suggest utilizing the repository connector to configurations. These connectors already request the metadata regarding the branches. We can request metadata concerning tags and check if there is a newer version in the metadata compared to the provided reftag. I proposed performing this in configuration.py during import and informing the user if there is a new version available.
Notes on the discussion based on this idea:
My goal is to inform users at import since some users won't check for a new version otherwise.
This idea does not completely solve all the issues above. Also, we might end up overlooking some fringe cases.
To cater to the api, such a check should have the option of being toggled off via and env var.
(Based on discussions on @xgui3783 and I, @AhmetNSimsek. @xgui3783 please feel free to edit if I missed sth, or comment new points.) Version deprecation could involve a few possible reasons:
neuroglancer.humanbrainproject.eudies or it needs to be migrated for another reason.I approached @xgui3783 with the following initial idea: For a simple check of the newest version, I suggest utilizing the repository connector to configurations. These connectors already request the metadata regarding the branches. We can request metadata concerning tags and check if there is a newer version in the metadata compared to the provided
reftag. I proposed performing this in configuration.py during import and informing the user if there is a new version available.Notes on the discussion based on this idea: