Dataset delta bundles 11 and 15 downloads for Norwegian Nynorsk do not work.
Just gives white page with an error:
NoSuchKeyThe specified key does not exist.No such object: common-voice-prod-prod-datasets/cv-corpus-15.0-delta-2023-09-08/cv-corpus-15.0-delta-2023-09-08-nn-NO.tar.gz
NoSuchKeyThe specified key does not exist.No such object: common-voice-prod-prod-datasets/cv-corpus-11.0-delta-2022-09-21/cv-corpus-11.0-delta-2022-09-21-nn-NO.tar.gz
Looking at the Deltas for v15 and v11 for nn-NO, I don’t think there’s any data in them. They are showing as 0 voices in the dataset download page. I think they’re empty, which would explain why this error is showing.
Perhaps we need an exception on the website to state that a delta file is empty, to prevent it being downloaded?
I says 1 validated hours for both:
Might miss out on some validated/invalidated data.
Having an archive with emtpy clips folder, but the tsv files might be better if there is any data there
There is already an issue filed in about this on github:
Actually they should be downloadable. The delta file is not empty, it should include the metadata, but because there are no new clips, it is not generated. The metadata is necessary for upgrading the previous version’s metadata, even if there are no new clips, some of the clips previously in other.tsv got reviewed.
I PM-talked with the team about that last week, and my general purpose upgrade script is waiting for a fix.
And IMO also deltas for v10, v11, and v12 are not usable for old + delta => new style workflows, the metadata is not correct. And IMHO they should be deleted from the database tables.
