@mattab do we need to take log deletion setting into consideration? If we invalidate a report, will the invalidated report eventually disappear or do we keep the report if there is no final report? Otherwise we risk some reports cannot be regenerated?

ping @mattab

To invalidate, we want to use the method ArchiveInvalidator::rememberToInvalidateArchivedReportsLater($idSite, $date)

We delete the dates based on log_visit.visit_last_action_time on the visit.

When we pass the $date to the method, need to make sure to apply the site's timezone. Otherwise we might invalidate the wrong date.

do we need to take log deletion setting into consideration? If we invalidate a report, will the invalidated report eventually disappear or do we keep the report if there is no final report? Otherwise we risk some reports cannot be regenerated?

AFAIK the API which invalidates reports checks when the first logs are available and if the logs are deleted for this date, the reports won't be invalidated. So we shouldn't have to worry about it.

AFAIK the API which invalidates reports checks when the first logs are available and if the logs are deleted for this date, the reports won't be invalidated. So we shouldn't have to worry about it.

so basically, if someone deletes a visit, and a user expects this data to be deleted from reports, it won't be actually deleted @mattab ? I don't think it's what users expect, at the same time deleting those reports is also not what people expect :)

My comment was unclear... What i meant is: the API which invalidates reports checks the date-time of the oldest logs and if the date to invalidate reports for is older than this date-time then the reports are not invalidated (because we know there is no logs to re-archive reports). Since when we delete a visit it means there are still logs for this day for sure, it should always result in an invalidation down the line.

@mattab that's clear. The problem is when you delete a visit of 12 months ago, and log deletion is set to 6 months... then the archive won't be invalidated/deleted but a user would expect the data to be removed from that archive.

It's an edge case though and don't really need to think about it. Be only an issue when deleting visits and for those visits log purging hasn't happened just yet.

Guess it's more like about deleting older reports that contain personal data but that's a different issue.