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

IRC log for #mojo, 2017-07-03

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

All times shown according to UTC.

Time Nick Message
01:13 jberger Mercury
01:13 jberger oh karjala_ isn't online?
01:13 jberger karjala ^^
01:15 jberger sri: I like the fd= option
01:15 jberger seems like if systemd provides it, that would be a nice way to deploy
01:53 VVelox joined #mojo
02:20 noganex joined #mojo
02:44 karjala The  video's online alright, jberger, but I wasn't sure I heard correctly.
02:48 pink_mist karjala: he meant the nick you used to ask the question "karjala_" had gone offline by the time he had a chance to answer the question
02:49 preaction but woo Mercury!
02:50 karjala oh. https://github.com/preaction/Mercury ?
02:50 karjala kewl
02:50 preaction yes
02:51 preaction i should give an actual talk/blog or something about it and stop treating it like an experiment that went too far ;)
02:51 preaction hmm... if i fatpack, i could give that talk at other conferences...
03:05 karjala dockerize!
03:06 karjala (maybe - I know nothing about docker)
03:07 preaction it's a possibility to be sure. mercury is simple enough that all you need to do is start it up
03:07 karjala I see one problem, though...
03:08 karjala If I have 10.000 rooms (each user may open one), then all hyupnotoad processes need to open 10.000 websocket connections, ,isn't that so?
03:09 karjala maybe add /sub-multi, ,and let the hypnotoad process add/remove rooms dynamically through the single connection?
03:09 preaction yes
03:09 preaction that's also a possibility
03:09 preaction you could also just have one websocket, with the room in the message, and ignore messages that aren't for the right room.
03:10 karjala yes, that too
03:10 karjala maybe more economical, too
03:11 karjala yes i'd follow your idea (if I had a running project that needed it)
03:11 preaction mercury is internally simple, and its api is message-agnostic (mercury does no parsing/understanding of the messages that pass through it). that necessitates certain things, and makes it suboptimal for the bit you mentioned. the other solutions have other pros/cons
03:11 karjala But one thing that I think is needed is this:
03:11 preaction i could provide a /multi api that did something a bit different, but now we're talking message framing and other fun stuff that's annoying. though now that you got me thinking, i might have a solution to it...
03:11 karjala oh, it's not needed sorry... I was going to say EVENTS, on connect, on disconnect
03:12 karjala but ua::websocket hanbndles that already
03:12 preaction yep!
03:12 karjala that's excellent... and the server's code is running C, you say?
03:13 karjala is C, I mena
03:13 karjala mean
03:13 preaction mercury is pure-Perl using Mojolicious
03:13 karjala o
03:13 preaction i only still consider it a toy because of how stupid-simple Mojolicious made it to create ;)
03:13 karjala lol ok
03:13 karjala Have you stress tested it to see how many messages/sec it can handle?
03:14 preaction first version was a hundred lines of code, tops
03:14 karjala Is it stable & reliable?
03:14 karjala how many msgs, before it drops random msgs, I mean
03:15 preaction http://preaction.me:3000 this has been online for at least a year now, without me watching it
03:15 karjala wow
03:15 preaction websockets are tcp, so there's no such thing as dropping
03:15 preaction i haven't done any stress-testing though
03:16 preaction i'm sure that'd be a fun project though: write a stress-test bot and run it through a profiler and optimize things
03:16 karjala Would it be possible to optimize bits of it by writing them in C with use Inline C;  ?
03:17 preaction the only optimizing, i suspect, would be in Mojolicious itself
03:17 preaction like, i don't think C is going to help much without having to just wholly rewrite the thing in C (and, if you're doing that, why not ZeroMQ or Nanomsg?)
03:18 karjala yes, ok
03:48 inokenty-w joined #mojo
05:25 kaare joined #mojo
05:42 karjala_ joined #mojo
06:45 AndrewIsh joined #mojo
06:49 dod joined #mojo
06:54 dod joined #mojo
07:02 Vandal joined #mojo
07:17 ashimema anyone got a link to the video of jberger doing a talk mentioning message brokers above.. I'm interested in watching that ;)
07:21 preaction that's the one on Vue.js, Mojolicious, and Chat or something
07:25 preaction https://www.youtube.com/watch?v=t9ZZq22pfag&index=77&list=PLA9_Hq3zhoFxdSVDA4v9Af3iutQxLI14m
07:30 sri minion would also be a great talk topic
07:31 sri it's crazy how mojo is basically on a monthly release schedule now, who would have thought that's possible two years ago :o
07:32 sri back then releases were almost daily
07:35 sri and we are free of deprecations again
07:35 good_news_everyon joined #mojo
07:35 good_news_everyon [mojo] kraih pushed 1 new commit to master: https://git.io/vQEwP
07:35 good_news_everyon mojo/master 4d9b7f3 Sebastian Riedel: remove deprecated watch attribute from Mojo::Server::Morbo
07:35 good_news_everyon left #mojo
07:47 dod joined #mojo
07:50 karjala_ joined #mojo
07:52 ashimema thankyou :)
08:27 trone joined #mojo
08:29 schelcj joined #mojo
08:51 schelcj joined #mojo
08:53 rshadow joined #mojo
09:00 schelcj joined #mojo
11:04 bc547 joined #mojo
11:15 absolut_todd joined #mojo
11:20 tchaves joined #mojo
11:23 rshadow joined #mojo
12:00 howitdo joined #mojo
12:44 Pyritic joined #mojo
13:10 jlaffaye_ left #mojo
13:14 jabberwok https://github.com/lindleyw/testers   a long way from done, but moving to Mojo and Minion makes this far easier to control and monitor
13:16 marty joined #mojo
13:16 brunoramos joined #mojo
13:25 maschine joined #mojo
13:33 gregf_ joined #mojo
13:36 marcus jabberwok: 404
13:37 pink_mist marcus: 0_o his link works fine for me
13:38 kwa marcus: Do you play "Hill Climb 2"? I raced against a "marcus perl" a few days ago, so unless it's a surname I'm not aware of...
13:39 marcus hmm, seems there's a convos bug :-o
13:40 marcus kwa: Not me :)
13:40 kwa Ah. :)
13:40 CandyAngel marcus: Too soon. You should ask who won first!
13:41 pink_mist lol
13:41 marcus CandyAngel: :D
13:41 pink_mist CandyAngel: maybe he remembers :P
13:41 marcus The North Remembers.
13:45 gryphon joined #mojo
14:01 gizmomathboy joined #mojo
14:11 sri hope someone submits a mojo talk for tpc amsterdam
14:11 sri we are already very badly represented in amsterdam :S
14:12 zen aw thats too bad, we are using mojo in both of our products and are very happy with it
14:14 zen our CTO doesnt get the futurama references though :P
14:14 sri time to get him a dvd box set :p
14:15 sri or her
14:15 zen didnt know about TPC though, looking into it (I live nearby Amsterdam)
14:16 ribasushi zen: so does sri ( for decent values of nearby ;)
14:17 sri about 4 hours by car
14:18 sirhalos joined #mojo
14:18 ribasushi hm indeed, mental map had you further west ( Muenster-ish )
14:18 ribasushi sri is thereby excused from attending ;D
14:19 sirhalos Hello, I have a Mojolicious app (not lite app), with several modules in the Controller dir. Now for my main routing module is there a way to not have to have all the routes in there? Instead I would prefer just a pointer to groups of routers in a different file.
14:20 sirhalos I'm referring to the sub startup
14:20 sri ribasushi: i'm between hamburg and bremen
14:21 zen sri: do you attend CCC?
14:21 ribasushi yeah I knew that but what I had in my head was "road from Aachen to Ams forms a triangle with where sri is", which is very far off
14:21 sri nope, i don't go to too many conferences
14:22 zen I was there 2 years ago, man it was crowded
14:22 ribasushi large conference is large
14:24 ribasushi zen: supposedly Messe Leipzig this year should be more size-appropriate
14:25 zen oh they finally made the move? great
14:25 zen im not claustrofobic but sometimes it really got way too crowded
14:25 ribasushi zen: CCH is no longer available - full building refurb underway
14:26 zen ahh
14:26 ribasushi there was a good chance of 34c3 not happening at all, the venue got finalized like a month ago
14:26 zen hmm
14:27 genio I don't remember the last time I've been to any conference. :/
15:03 jberger sirhalos: not really, we favor explicit routing in Mojolicious
15:03 jberger there are some shortcuts that can make the routing table easier on the eyes
15:03 marcus sri: Didn't submit a talk, but both me and batman will be at tpc amsterdam, (and the rest of reisegiganten devs). Will probably do a related lightning talk tho.
15:04 sirhalos Okay, my problem is the amount of routes I have in my Mojo::Base 'Mojolicious'; startup sub is getting long.
15:04 jberger marcus / batman: you want to run my mojolicious introduction talk?
15:04 jberger already written, ready to go
15:04 jberger http://jberger.github.io/MojoliciousIntroduction/#/
15:04 jberger (did that paste twice? strange)
15:05 jberger sirhalos: you might make methods that establish routes to break them up
15:06 jberger sub setup_user_routes { ... } sub setup_resource_routes { ... } sub startup { ...; $app->setup_user_routes; $app->setup_resource_routes; ... }
15:07 sirhalos Okay I think I get what you mean. I'll try that. Thanks.
15:18 jberger sigh, I just saw that my Mojo talk was posted and they messed up the title bar
15:18 jberger several of them have been messed up
15:18 jberger and when I've commented on the videos they've been fixed
15:18 jberger but why are they even there? they just take up room on the screen
15:18 jberger you know what presentation you are watching surely
15:19 pink_mist not sure I get what you're talking about? link?
15:19 jberger https://www.youtube.com/watch?v=t9ZZq22pfag
15:19 trwww1 joined #mojo
15:20 pink_mist is it the "Chris Prather - "Make New Friends"" thing you're talking about? yeah, that's weird
15:20 jberger it isn't the only one that has been incorrect
15:20 jberger and like I said, they've been good about correcting them once pointed out
15:20 pink_mist right
15:20 jberger but it already has ~100 views
15:21 jberger which will be lost
15:21 jberger if they correct it
15:21 jberger and I don't even get why the title bar is there
15:21 pink_mist I guess I shouldn't watch this version then; save my view so it gets counted :P
15:21 jberger I guess maybe just to take up space for having both screens
15:21 jberger meh, watch it now if you want, one view +/- doesn't matter
15:22 pink_mist they could have used a nicer colour for it I haveto say ... that purple one is really an eyesore
15:22 jberger overall though, in the "perl is/isn't dead" debate, we are just throwing away views
15:22 pink_mist pink would have been much nicer :P
15:23 karjala_ left #mojo
15:29 pink_mist jberger: the "Make New Friends" bit is a bit apropos though, since you make the audience chat with eachother :P
15:29 jberger that is true
15:31 zivester joined #mojo
15:37 bjakubski joined #mojo
15:41 ptolemarch joined #mojo
15:42 pink_mist jberger: nice talk =)
15:42 jberger thanks
16:07 sh14 joined #mojo
16:16 sri marcus: ah, that's cool
16:22 pink_mist sri: is suse sending anybody to tpc amsterdam? if so, why aren't they sending you? :P
16:22 pink_mist if not, why
16:22 pink_mist or are they sending you?
16:25 sri pink_mist: suse would send me to conferences if i gave a talk related to suse
16:27 sri they are not usually sending people to perl conferences
16:28 pink_mist maybe you can get them to change that :P
16:29 sri i could prolly get away with talking about the mojo features i added for my work at suse ;p
16:29 sri but it would have to be planned a little more in advance
16:29 pink_mist yeah, understandable =)
16:31 sri programming language communities are not really a big thing in general at suse, it's all more about linux and cloud
16:35 sri in related news, coolo found a bug in a work project that turned out to be another bug in our graceful worker shutdown code
16:36 sri so we still have situations where connections get closed before a request has been processed
16:37 marty joined #mojo
16:37 sri yea... soo if you complained about that recently and i said that's impossible... well... oops... my bad :)
16:38 sri will test and fix it tomorrow
16:39 sri it's Mojo::Server::Daemon::close_idle_connections most likely, if someone want to beat me to it
16:39 marty joined #mojo
16:49 marty joined #mojo
17:01 sirhalos Moving over from mojo lite to mojo regular and all my get routes are being seen and work from my control module but by my post and put are not. Here is an example of my put $routes_ref->put('/sheepdog/system/create/:type/:name')->( controller => 'System::Bovid', action => 'create' );
17:02 Grinnz that second part should be a ->to method
17:03 sirhalos the action => 'create' is the method found in controller System::Bovid
17:04 pink_mist sirhalos: yes, but you did ->( controller => ... ); ... you should have done ->to( controller => ... );
17:05 sirhalos still didn't work :(
17:05 sri put routes are no different than get routes, aside from the name
17:06 sirhalos yeah that's why I'm confused, my gets work fine and show up with routes -v
17:06 sirhalos and the put and post worked in mojo lite
17:06 sri if they are not in routes -v, they don't exist, at all
17:06 sri even shitty borked routes show up in routes -v
17:11 jberger sirhalos: are you chaining the creation of the POST from the creation of the GET?
17:12 sirhalos you mean in my restclient tool?
17:14 jberger no, in your startup method
17:14 jberger if you chain a POST after a GET then nothing will be able to route to it
17:14 sirhalos https://pastebin.com/EgVcgiuq
17:16 sirhalos you mean I need to move all my post and put before any of my get's in the startup?
17:16 jberger no
17:17 jberger that routing table looks fine
17:17 sirhalos line 78 and 85 have my new post and put's
17:18 jberger and they do or do not show up in routes -v?
17:19 sirhalos They do not, only my GETs
17:20 jberger that makes absolutely no snse
17:20 jberger sense
17:22 sh14|2 joined #mojo
17:23 sirhalos Here is my output from routes -v https://pastebin.com/PSGvCYyR
17:25 jberger when I run it I see the POST and PUT
17:25 jberger are you sure you are using the correct application script
17:26 jberger err, application class file
17:26 jberger test it out, comment out one of the routes that you see there
17:26 jberger then check the routes again
17:26 jberger I bet you still see it
17:26 jberger I bet you aren't editing the application class file that is being used
17:31 jberger once you've confirmed that (I'm 95% sure you will) I have a few other comments, if you are interested
17:36 sirhalos I see them now. I'm so sorry, I was editing a copy and the copy I was running was on a different server. I feel like such a noob.
17:36 sirhalos I'm so sorry Joel to waste your time on that.
17:36 PryMar56 joined #mojo
17:37 sirhalos But I would be interested in any of your expert comments, of course.
17:41 jberger its ok, we all do it
17:42 jberger first big question, are those routes structured in a way that you can't change them yet?
17:42 jberger it is much more RESTful to have /sheepdog/system/bovid/:bovid/file etc rather than how you have them
17:42 sirhalos no I can change them, its still in testing no applications have been made around them.
17:43 sirhalos Okay, i wasn't sure how to do them, I originally had the old CGI style but discovered I should change that.
17:43 jberger give me a sec, let me restructure your route tree, I think you'll like this style better
17:54 jberger sirhalos: http://paste.debian.net/974583/
17:54 jberger I didn't get all the way through it, but I think you see where I was going
17:55 jberger heh, my changes are pretty easy to see given our style differences :-P
17:55 sirhalos oh yes, that looks much better. A lot less copy and paste code.
17:59 jberger also shows much easier that it is a routing tree
18:05 tchaves joined #mojo
18:13 tchaves joined #mojo
18:40 suede joined #mojo
18:55 Lee[home] jberger: my TPC talk just got reuploaded too (no idea why)
18:56 FROGGS joined #mojo
18:59 dod joined #mojo
19:32 marcus sri: Very interesting about the graceful workers, could explain spurious 500s we've been seeing on upgrading.
19:34 jberger Lee[home]: I reported yours, the slides and the A/V were out of sync
19:35 jberger by a lot
19:38 Lee[home] ah!
19:38 Lee[home] thanks!
20:22 sri marcus: yea, that explains it
21:04 ferreira joined #mojo
21:08 ferreira I am curious about the mismatch between behavior to signals among Hypnotoad and minion workers
21:09 ferreira https://metacpan.org/pod/distribution/Mojolicious/lib/Mojo/Server/Hypnotoad.pm#MANAGER-SIGNALS vs https://metacpan.org/pod/Minion::Command::minion::worker#SIGNALS
21:09 ferreira INT/TERM stops gracefull Minion workers, while QUIT stop them immediately. And the reverse is true for Hypnotoad.
21:09 ferreira Are these different convention for HTTP servers vs queue workers? or something that slipped by and got frozen due to backcompabibility?
21:12 Grinnz they have different requirements
22:02 Peppard joined #mojo
22:37 VVelox joined #mojo
23:29 schelcj joined #mojo
23:53 gizmomathboy joined #mojo

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