[01:57:42] * ebernhardson wonders if deduplicatable is too awkward....probably :P
[02:27:34] <ebernhardson>	 the more i think about this deduplication...now i'm wondering why we can't just emit the last event of the window
[02:28:06] <ebernhardson>	 We only have two possible actions: delete the page, replace the page with current cirrusdoc. These are inverses of each other, so it never makes sense to emit more than one
[02:29:06] <ebernhardson>	 which one wins? I can't come up with any reasonable answer other than the one with the most recent timestamp
[05:15:35] <ryankemper>	 @in.flatador: Re-image of 2021 looks happy. I set a downtime on the host. Haven't kicked off the data transfer; we should get that going tomorrow morning from 2022->2021
[08:18:31] <dcausse>	 ebernhardson: indeed, on the same rev_id I don't see other ways to discriminate than using the event time
[08:19:20] <dcausse>	 and indeed "deduplicatable" sounds a bit too weird :P 
[08:20:39] <dcausse>	 the french alternative sounds a bit better "déduplicable", but only for french I suppose :)
[09:48:52] <gehel>	 lunch
[10:38:02] <dcausse>	 lunch
[12:42:46] <inflatador>	 <o/
[12:45:20] <dcausse>	 o/
[13:30:57] <inflatador>	 maybe "duplicated" instead of "deduplicatable"? Not sure if that conveys the concept well enough
[13:31:09] <ebernhardson>	 \o
[13:31:28] <ebernhardson>	 it's slightly different, deduplicated is after the deduplication has occured. Deduplicable is the state is being able to be deduplicated
[13:34:08] <dcausse>	 o/
[14:47:02] <inflatador>	 <o/
[14:56:34] <inflatador>	 maybe 'dedupable' ? 
[16:01:23] <inflatador>	 workout, back in ~40
[16:39:35] <inflatador>	 back
[17:44:39] <inflatador>	 lunch, back in ~45
[18:30:46] <inflatador>	 back
[20:30:45] <ebernhardson>	 meh, nothing is ever simple :P  Now thinking about what exactly happens if we pass-through late events and let them skip deduplicate and merge. 
[20:30:47] <ebernhardson>	 t
[20:38:31] <ebernhardson>	 so for example if we receive a delete then an undelete, but the delete is late, then they process out of order and the final state is the delete wins :S
[23:11:32] <ebernhardson>	 intellij suggests the best names. inputEventInputEventOneInputStreamOperatorTestHarness