When a publisher stops publishing totally, we should purge all data

[odscjames]

This follows on from #18

[robredpath]

Delete all data from a publisher that dissapears

[robredpath]

Establish what we need to do if a publisher stops publishing either a particular feed or OA data entirely

[robredpath]

Discussed with @thill-odi . We shouldn't actively delete data from a publisher that disappears, but we shouldn't seek to retain it (eg if we need to wipe the db for some reason) either.

The exception is if there's a takedown request, which we'll deal with separately.

[thill-odi]

Sorry, looking at https://blog.openactive.io/privacy-by-design-what-gdpr-means-for-your-open-opportunity-data-b1f5a805e1e6, it seems we will indeed need to do a wipe of everything older than [insert reasonable amount of time here], regardless of whether the data feed is in fact up or not.

I'll confer with the author of that blog post regarding what a reasonable amount of time would be and note that here.

[thill-odi]

Update: though it would be possible to split this hair quite finely, for present purposes let's keep this simple and purge all data relating to events >90 days in the past.

[robredpath]

@thill-odi This may be entirely missing the point, but is there a risk that someone publishes some far-future events, then unpublishes them, but we continue to display them because they relate to things that are in the future, still?

[thill-odi]

This is an edge case. The most common publishing scenario is two-weeks-in-advance. Should somebody publish further in advance and then cancel these future events, to be in conformance to the spec they can't just delete the events from their feed: they need to indicate the fact of deletion in their feed, which should then flow through to any harvesting that's done and result in deletion there.