Open iulianacse opened 1 year ago
We determine if a version is auto approved, rather than manually approved, by the existence of an AutoApprovalSummary
for that version, with a verdict of AUTO_APPROVED
. But it seems we didn't start creating the AutoApprovalSummary
instances for unlisted versions until 2019-11-21 (listed started 2 years prior) so it's an unreliable method for old versions.
@eviljeff Indeed, it is reproducible with https://reviewers.addons.allizom.org/en-US/reviewers/review-unlisted/619823 on AMO stage
@wagnerand can you comment here how important you think this is - it's only effecting unlisted versions that are 2 1/2 years old (and technically listed versions more than 5 years old, if we still have any of those that are valid)
@eviljeff 5 years is a long time, but 2.5 is not that long. It would be nice if we could address it, but I am not sure we even can? Since there is no AutoApprovalSummary
, how would we know? Or would we just create fake, mostly empty ones?
@eviljeff 5 years is a long time, but 2.5 is not that long. It would be nice if we could address it, but I am not sure we even can? Since there is no
AutoApprovalSummary
, how would we know? Or would we just create fake, mostly empty ones?
My first thought was to just change the label to be ambiguous, but if you want to actually confirm the auto approval we'll need to backfill with fake-ish AutoApprovalSummary
instances (the confirmation is stored on the AutoApprovalSummary
).
It's not super trivial to backfill - we think it was always signed in a task (so we can filter approval activity logs based from the task user) but it's old code (deleted in 2019) but there are some unknowns.
Old Jira Ticket: https://mozilla-hub.atlassian.net/browse/ADDSRV-469
Describe the problem and steps to reproduce it:
After reviewing an unlisted add-on, when I wanted to confirm approval on the last versions I noticed that all versions in the list appear to have been manually approved, but no versions have been actually manually approved.
What did you expect to happen?
Versions that have not been manually approved should not appear in the list as such.
Anything else we should know?
Add-on link: https://reviewers.addons.mozilla.org/en-US/reviewers/review-unlisted/903037 In the screenshot bellow you can see the list, where version 2.0.14 appears to have been Manually Approved In the next screenshot we can see that version 2.0.14 was approved only by the system.
┆Issue is synchronized with this Jira Task