Duplicate docket entries (1 from rss?)

[johnhawkinson]

In this case, https://www.courtlistener.com/docket/6296809/calderon-jimenez-v-cronen/?page=2, it appears that RECAP has two entries for the same order. Presumably one from RSS and one from the docket report and they did not merge properly:

"Huh."

[mlissner]

Hm, somehow we got that as an unnumbered entry though it seems to be a numbered entry now. That's pretty weird.

[johnhawkinson]

interim conclusion: we need to save raw rss for debugging. Maybe for other things too.

[mlissner]

Yeah, wouldn't hurt. We can open a bug for it, but I have no idea when I'll get to it.

[johnhawkinson]

This is not a one-off. Today's example: https://www.courtlistener.com/docket/6151669/alasaad-v-duke/?filed_after=&filed_before=&entry_gte=&entry_lte=&order_by=desc

[johnhawkinson]

And then there's this peculiar variant, where it looks a "Notice of Consent - No Consent" was edited to "All parties have consented" but both appear in the RECAP docket report, one unnumbered. I wonder if this means if we had reparsed the RSS feed after the edit (e.g. w/i 24 hours), we would have gotten the proper information.

https://www.courtlistener.com/docket/6106337/dalessio-v-university-of-washington/?filed_after=&filed_before=&entry_gte=&entry_lte=&order_by=desc

So the duplication is a bigger problem in the face of edits.

[ikeboy]

Still an issue, e.g. https://www.courtlistener.com/docket/16554647/infinity-global-consulting-group-inc-v-tilray-inc/

[johnhawkinson]

Yes, this is, unfortunately, a regular problem. It's pretty confusing, we should prioritize fixing it, and then somehow managing to repair the months of busted data it has lead to :(

[mlissner]

Is there a way to fix it? The only thing these items have in common, I think, is the date?

[johnhawkinson]

Oh, it's certainly fixable. This problem was introduced when changes were made with respect to de_seqno and ordering, or something like that. I thought we had a conversation around the change that introduced this bug, perhaps in the Slack. In any event, it was shortly prior to my opening this Issue, so probably Feb or March 2019.