[10:02:52] <gehel>	 lunch
[10:04:23] <dcausse>	 lunch 2
[11:59:20] <pfischer>	 lunch + errands
[15:02:37] <gehel>	 dcausse: triage in https://meet.google.com/eki-rafx-cxi
[17:10:20] <dcausse>	 dinner
[17:39:09] <ebernhardson>	 hmm, airflow-dags tests pass locally....but CI is failing with `libssl.so.3: cannot open shared object file: No such file or directory` :(
[18:17:51] <inflatador>	 <o/
[18:40:01] <ebernhardson>	 gitlab really doesn't make it easy to recreate what CI is doing locally to debug :(
[21:24:38] <inflatador>	 Lunch/aquarium, back in  ~2h
[21:37:57] <ebernhardson>	 flink docs on late events are dreadfully unclear, going to have to do some testing.
[21:44:26] <ottomata>	 https://nightlies.apache.org/flink/flink-docs-master/docs/dev/datastream/operators/windows/#allowed-lateness
[21:44:27] <ottomata>	 ?
[22:07:04] <ebernhardson>	 ottomata: right, but the question is how the timing of it all works out. I think what happens, but it's not clear, is that events are "late" but allowed into the window, which results in multi-fires of the same window, but then events that come if after the allowed lateness can be redirected to a side output
[22:07:29] <ebernhardson>	 but it's not particularly clear, i find the use of "late" in both context's with a slightly different definition to be imprecise at best
[22:08:12] <ebernhardson>	 a late event results in multiple fires of the window, and a late event goes into the late events side output. but the second late is after the first late :P