02:46 <daveg[m]> I have run into a problem with Kernel#to_enum vs #each, and which I think is also tied to JRuby version. It's also unfortunately tied to ActiveRecord (4.2.11 -- upgrading that happens after i finish upgrading JRuby). https://gist.github.com/dgolombek/3123d5602ff6d76e96f2bac01b570210 is the example, but it's not self-contained -- you need an AR class and a way to construct it. All the paths work in MRI 2.6.6. In this toy

02:46 <daveg[m]> example, I see consistent behavior between JRuby 9.1.17.0 and 9.2.11.1, but when in a larger system, I'm seeing different behavior (9.1.17.0 works, 9.2.11.1 fails with Enumerator#peek hanging). I'll submit an Issue RSN but wanted to mention it here in case you have any suggestions for better ways to set up a spec that depends on ActiveRecord. Thanks!

03:37 <headius[m]> hmmm

03:38 <headius[m]> dave.g: you could test 9.2.12 but I doubt it fixes it... you should open an issue

03:38 <headius[m]> we may be able to diagnose without reproducing, if we can get some thread stack dumps from you

03:39 <headius[m]> Enumerator#next and peek and friends spin up a thread to "drive" the iteration, so there could be some conflict there like a deadlock

03:39 <headius[m]> in any case open an issue and we can get to the bottom of it

08:30 <headius[m]> kares: is there any automation for updating the jdbc gems?

08:50 <headius[m]> kares: I see the jars are just versioned in the repo... updated to 42.2.14 to address CVE-2020-13692 and pushed a new gem

09:19 <kares[m]> no automation just cp-ing jars over s you figured ...

15:05 <enebo[m]> kares: didn't we try to update one of those jars in the past and backed off due to some issue?

15:18 <kares[m]> enebo: yes

15:18 <kares[m]> there were CI failures for AR 5.X something, I do not remember

15:18 <kares[m]> hopefully the postgresql adapter is looking good ...

15:18 <daveg[m]> Release notes link on https://www.jruby.org/download still points to 9.2.11.1 release notes, instead of 9.2.12.0

15:20 <kares[m]> it run 🟢 so at least recent 6X should be covered, not sure about older 5X

15:20 <enebo[m]> dave.g: thanks...will fix

15:20 <kares[m]> * master run 🟢 so at least recent 6X should be covered, not sure about older 5X

15:20 <enebo[m]> kares: ok I could not remember which one it was but I thought it was postgres

15:22 <enebo[m]> dave.g: updated...I will check in a few to make sure it redeployed properly

15:23 <enebo[m]> out downloads page is some hard-coded html file for reasons I cannot remember and not used by our config. Feels like we should revisit our generation tech :)

15:25 <enebo[m]> link is fixed

15:27 <daveg[m]> Great, thx

19:48 <headius[m]> enebo: you make any progress with that Enumerator issue?

19:48 <headius[m]> I see you were poking at it

19:49 <enebo[m]> headius: not with having to make a database and install rails

19:49 <headius[m]> ah, sure

19:49 <enebo[m]> although I don't get that since it uses factory_bot

19:50 <enebo[m]> So we still have nothing we can reproduce without some work

19:50 <enebo[m]> I was hoping to get it triaged but it is still lacking

19:50 <headius[m]> dave.g: so it seems to work ok with .each enumerator?

19:51 <headius[m]> I see your example spec seems to claim that .each works but .to_enum doesn't

19:51 <enebo[m]> aha I did not notice dave.g was the reporter

19:52 <headius[m]> dave.g: a full trace would be helpful... get it to hang again and then ctrl+\ or send use `jstack` command to get a JVM stack trace

19:52 <headius[m]> I'll post in bug too