The web in a box - a next generation web framework for the Perl programming language

IRC log for #mojo, 2017-03-03

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:16 stryx` joined #mojo
00:29 Janos joined #mojo
01:03 aborazmeh joined #mojo
03:18 noganex joined #mojo
03:48 asarch joined #mojo
04:13 lluad joined #mojo
05:04 dboehmer_ joined #mojo
05:38 inokenty-w joined #mojo
05:40 saki joined #mojo
06:00 arhak joined #mojo
06:32 stryx` joined #mojo
06:40 polettix joined #mojo
07:22 kaare joined #mojo
07:34 janl joined #mojo
07:36 polettix joined #mojo
07:40 dod joined #mojo
07:48 Vandal joined #mojo
08:04 sri jberger: we demand updates!
08:11 AndrewIsh joined #mojo
08:39 trone joined #mojo
09:03 rshadow joined #mojo
09:59 prg joined #mojo
10:00 sugar joined #mojo
10:53 prg quick minion question: it's ok to call $job->finish() or $job->fail() inside a task, right?
10:54 sri yes, that's how you set a result value
10:54 polettix joined #mojo
10:55 prg yeah, thought so. Minion::Notifier is giving me trouble if i do that, and i'm having a hard time pinpointing the cause. jberger, you around?
10:57 nic shouldn't really ask personal questions
10:57 nic we try to be non-shapist here
10:57 prg funny.
10:58 prg let me rephrase that
10:58 prg jberger, do you have some spare time to help me debug a Minion::Notifier problem?
10:59 prg maybe it's not even Minion::Notifier related, might be a Mojo::Pg problem...
11:00 nic Are you able to consturct a mini-example that illustrates the issue?
11:01 prg yeah, let me pastebin that
11:02 prg http://pastebin.com/r0Bt8Fvm
11:03 prg if i run that as "script/myapp get /foo", jobs are enqueued as i'd expect
11:03 prg however, if i run a worker afterwards, eg "script/myapp minion worker" it bails intermittently with stuff like "message type 0x49 arrived from server while idle"
11:04 prg when i replace "$job->finish()" with a "return" this doesn't happen
11:05 prg oh yeah, i also get "DBD::Pg::db do warning:  at <...>/5.24.0/Mojo/Pg/Database.pm line 59."
11:06 prg if i remove "plugin 'Minion::Notifier'", $job->finish() works fine, too
11:09 sugar joined #mojo
11:25 prg i guess it could also be postgresql's fault, although it's pretty much a standard installation (9.5 from ubuntu 16.04)
11:26 marty_ joined #mojo
11:29 itaipu joined #mojo
11:29 prg probably not helpful, but postgresql's log says: "DEBUG:  unexpected EOF on client connection"
11:56 kes joined #mojo
11:56 kes Hi. How to configure hearbeat_timeout from startup?
11:56 kes http://mojolicious.org/perldoc/Mojo/Server/Prefork#heartbeat_timeout
12:17 kes I have stand alone script which I run from controller action
12:18 kes The object created at this script:
12:18 kes has ua       =>  sub{ Mojo::UserAgent->new };
12:19 kes It seems this ua is not attached to IO::Loop
12:19 kes Because I every time get: Worker 99 has no heartbeat, restarting
12:19 kes error
12:20 kes despite on increased timeout:
12:20 kes $self->ua->inactivity_timeout( 120 );
12:21 kes This occur only when I call this script's method from controller action. If I run this from command line it works fine
12:31 kes it seems like while I process request in controller I can not send another
12:41 kes How can I debug what is happening with Mojo::UserAgent at this moment?
13:02 sugar joined #mojo
13:16 kes joined #mojo
13:35 gryphon joined #mojo
13:43 tchaves joined #mojo
13:43 tchaves1 joined #mojo
13:50 pink_mist kes: you're supposed to use the $c->ua for that
13:50 pink_mist hmm, maybe
13:51 kes actually it is
13:51 kes http://paste.scsys.co.uk/555981
13:51 kes line 14
13:54 pink_mist well that ua will be attached to the loop; that's not the ua you described at the start though
14:18 bwf joined #mojo
14:20 gizmomathboy joined #mojo
14:23 polettix joined #mojo
14:25 sugar joined #mojo
14:29 jberger prg I just woke up, I'll try to take a look later
14:29 jberger The warning you get is a little troubling
14:30 jberger It's a warning that I used to get when doing very strange forking things with postgres notify
14:30 jberger I hope nothing has changed to make that happen in more normal circumstances
14:35 Janos joined #mojo
14:42 Pyritic joined #mojo
14:49 prg jberger: thanks, no hurry
14:49 prg i opened an issue, we can continue this there
15:08 PopeFelix joined #mojo
15:12 polettix joined #mojo
15:32 itaipu joined #mojo
16:05 PryMar56 joined #mojo
16:13 genio sri: Does Mojo::URL order parameters in a specific way? (https://github.com/libwww-perl/URI/issues/35)
16:15 sri genio: of course, order matters
16:15 sri Mojo::Parameters is entirely built around that
16:16 itaipu joined #mojo
16:16 lluad joined #mojo
16:16 genio So, I should likely update URI to order the parameters in the same way as Mojo::Parameters
16:19 sri if you change order the query strings become semantically different according to the specs
16:19 sri which may or may not be important, depending on the apps involved
16:21 genio I'll have to spend a little bit getting up-to-speed on how URI does parameters.
16:22 genio sri: Thanks!
16:44 sugar joined #mojo
16:55 sugar joined #mojo
17:10 sh14 joined #mojo
17:21 prg jberger: Minion::Notifier is probably not the culprit, but i'm still lost. i'd appreciate your input on this, see my last comment on the issue. heading out soon to see my newborn nephew, i'll be back monday
17:21 jberger prg: yeah, I'll try to replicate at some point today
17:22 jberger the fact that you get the "message type 0xHH arrived from server while idle" warnings is worrisome
17:23 jberger I've only seen those in situations where forking got weird
17:24 Pyritic joined #mojo
17:24 jberger I'll admit that though this module was very important at my last job, I don't use it at my current one, so it is possible that something has crept up and I didn't notice
17:25 jberger prg: can you try running this test? https://github.com/jberger/Minion-Notifier/blob/master/t/pg.t
17:25 jberger obviously it doesn't run automatically on installation
17:25 jberger set TEST_ONLINE_PG=postgresql://... in your environment
17:26 jberger (erg, I see that the notice about that is wrong :( )
17:27 prg hold on
17:27 prg All tests successful :/
17:28 jberger hmmmm
17:28 jberger do you have EV installed?
17:28 jberger the test forces the Poll reactor
17:28 prg well, doesn't surprise me. your test tasks don't call $job->finish
17:28 rshadow joined #mojo
17:28 prg nope, no EV
17:29 jberger oh, I didn't notice that ->finish was required from inside the job
17:29 jberger that is interesting
17:30 jberger sorry, I have dived in to the ticket yet
17:30 jberger haven't
17:30 prg no sweat
17:30 jberger I just noticed those warnings that have plagued me before
17:30 prg i just changed pg.t to use ->finish/->fail
17:30 prg now it fails
17:31 jberger can you do the same thing in the websocket test
17:31 jberger ?
17:32 jberger anyway, I now have a different suspicion about what is going no
17:32 jberger on
17:33 prg sorry, gotta run now. maybe i can make some time this weekend, otherwise it'll have to wait til monday
17:33 prg see ya
17:33 jberger o/
17:40 dod joined #mojo
17:58 PopeFelix joined #mojo
18:15 zivester joined #mojo
18:36 sugar joined #mojo
18:36 genio sri: It seems that ticket was actually resolved quite some time ago but nobody closed the ticket.  Thanks for the input though!
18:52 jberger Your occasional reminder that when opening a bug report (and especially if it asks for it) please include the version of the library you are reporting on
18:52 jberger https://github.com/kraih/mojo/issues/1065
18:55 pink_mist jberger: I'd add: additionally, if you use the included perldoc pages to read the documentation instead of using the website, you'll get the documentation for the version you have rather than the latest one
18:55 jberger honestly, I'd almost feel condescending posting that
18:55 jberger that should be obvious IMO
18:56 pink_mist I think you'd be surprised
19:18 haarg i like that he saw the VERSION HERE thing and thought "nah fuck that it don't matter"
19:22 preaction they tried upgrading to see if they were out-of-date, they just didn't realize that wasn't going to work for an upgrade
19:31 jberger argh, I hate that distribution installers report that an unversioned module reports it as being up to date
19:31 jberger the root problem here is that they were depending on Mojo::UserAgent not Mojolicious and then when I said they should depend on 7.13 it still said they were up to date
19:35 preaction the other possibility is that they always assume it's out-of-date. even though i'm sure cpanm keeps metadata around and could actually check what dist a module came from...
19:38 Grinnz dist versions don't matter, since they could do a completely different thing from the module versions. as far as i know cpanm and cpan-outdated specifically check if the installed version is less than the indexed version
19:40 Grinnz this is the drawback of only versioning the main module
19:52 sugar joined #mojo
20:20 stryx` joined #mojo
21:20 sugar joined #mojo
21:59 marty joined #mojo
21:59 mcsnolte joined #mojo
22:34 marty_ joined #mojo
23:02 stryx` joined #mojo
23:21 marty_ joined #mojo
23:52 lluad joined #mojo

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary