Discussion: TimeTravel API is unreliable and should not be used

[ruebot]

The TimeTravel API which is used in this method to create a reply URL link is unreliable.

This thread on the Memento Development list, started in November 2018, raised the issue initially. The recent TravisCI test runs have identified that the issue has now cascaded to our project, which has been raised in the mailing list thread.

So far I believe that we have two options:

1. Remove the functionality until the problem is fixed.
2. Use another service, which ultimately means running and maintaining our own service for this functionality.

[ruebot]

Still getting hit with this a lot. Not in the test anymore. I imagine that is a specific removal from their caching system. However, the warclight demo site seems to really suffer a lot from it. We get lots of false positives for reply urls pointing to the UKWA's PyWb instance (:wave: @anjackson).

This entire collection comes from sites crawled from a few of the WAHR projects, and playback is all provided via the York University OpenWayback instance.

Looking at slide #25 of Martin's presentation, I wonder if that is because of the large number of false positives coming from the York instance which is an out-of-the-box OpenWayback instance, we some UI modifications.

I wonder if Martin on GitHub?

[ruebot]

Here are a couple examples:

[anjackson]

Unfortunately, this does seem to be a problem with the UKWA pywb installation. The TimeGate isn't behaving correctly as it seems to have been broken by the customisations we have in place: ukwa/ukwa-pywb#37

[ruebot]

So, if I'm understanding it correctly now, it's not actually the fault of the Time-Travel API? The finger pointing moves again, and I feel silly? 🤕

[anjackson]

Not as silly as I feel, as it turns out to be our fault.

[ruebot]

Well, I'll close this, and stand in solidarity of silliness with you 😄