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

IRC log for #mojo, 2016-01-22

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

All times shown according to UTC.

Time Nick Message
00:09 binlei joined #mojo
00:12 ZoffixWin joined #mojo
00:18 berov1 left #mojo
00:30 bwf joined #mojo
00:47 binlei joined #mojo
00:51 go|dfish joined #mojo
00:54 binlei joined #mojo
01:05 zivester joined #mojo
02:24 Averna joined #mojo
02:58 ZoffixWin left #mojo
03:45 noganex joined #mojo
03:52 binlei joined #mojo
04:17 c--__ joined #mojo
04:34 inokenty-w joined #mojo
05:30 binlei joined #mojo
05:56 Shaeto joined #mojo
05:58 binlei joined #mojo
06:01 binlei joined #mojo
06:04 binlei joined #mojo
06:06 binlei joined #mojo
06:09 binlei joined #mojo
06:13 binlei joined #mojo
06:27 panshin joined #mojo
06:27 binlei joined #mojo
06:37 binary joined #mojo
06:45 binlei joined #mojo
06:46 dod joined #mojo
06:49 binlei joined #mojo
06:52 dod joined #mojo
06:58 panshin joined #mojo
07:16 cpan_mojo Mojolicious-Plugin-FontAwesome4-4.5000 by JHTHORSEN https://metacpan.org/release/JHTHORSEN/Mojolicious-Plugin-FontAwesome4-4.5000
07:30 panshin joined #mojo
07:32 dod1 joined #mojo
07:35 sugar joined #mojo
07:51 panshin joined #mojo
08:12 Vandal joined #mojo
08:24 panshin joined #mojo
08:26 AndrewIsh joined #mojo
08:27 trone joined #mojo
08:41 osfabibisi joined #mojo
09:04 binlei joined #mojo
09:17 jontaylor joined #mojo
09:22 sue joined #mojo
09:31 panshin joined #mojo
09:38 berov joined #mojo
09:56 dod joined #mojo
10:12 binlei joined #mojo
10:36 mib_926az2 joined #mojo
10:36 mib_926az2 Hello
10:37 pink_mist good morning
10:38 mib_926az2 just wondering if someone could explain to me if Mojo will built the html for me round my perl scripts.  I am hoping that I do not have to remember HTML albeit happy to learn Mojo
10:38 mib_926az2 just before I install it, if not then maybe cgi might be the answer if I have to do HTML again
10:39 dod1 joined #mojo
10:39 pink_mist you can use taghelpers to write html for you
10:39 pink_mist http://mojolicious.org/perldoc/Mojolicious/Plugin/TagHelpers
10:40 pink_mist you will likely still need *some* html knowledge though
10:40 pink_mist also, going to CGI.pm is a _really_ bad idea
10:41 pink_mist not only is it slow as crap, it's also unmaintained and unmaintainable as well as bloated, bugridden, and insecure
10:41 pink_mist I'd stay far far away if I were you
10:42 Lee please don't use CGI.pm for HTML generation
10:42 Lee also, please don't use CGI.pm
10:42 pink_mist mib_926az2: we had report in here of one user who went from CGI.pm to Mojolicious, and his code ran 150 times faster
10:44 nic and he became happier and more successful
10:44 mib_926az2 wow thanks very much for this information
10:44 mib_926az2 what about form creations boxes dropdowns
10:44 mib_926az2 etc
10:44 nic mib_926az2: don't worry about the html -- it actually becomes fun with Mojolicious
10:44 pink_mist those are in the link I gave
10:45 mib_926az2 so I can create a basic frame for my script happy to do that good news that it becomes fun as HTMl does my head in
10:45 mib_926az2 will Mojo run on a mac
10:45 nic yes
10:45 pink_mist the lead developer uses a mac, so yes :P
10:45 mib_926az2 even better this weekend just got fun
10:46 mib_926az2 is there a front door to get started
10:46 pink_mist http://mojolicious.org/perldoc/Mojolicious/Guides
10:46 mib_926az2 thanks
10:47 pink_mist also, though they're a little outdated nowadays, these are awesome: http://mojocasts.com/e1
10:47 pink_mist https://github.com/kraih/mojo/wiki#screencasts <-- see here for list of things that have changed since the mojocasts were made
10:47 mib_926az2 stupid question for this forum but and honest answer I am sure, what is better or the differences between Dancer Catalyst and Mojo Padre etc
10:48 pink_mist mib_926az2: check here: http://shadow.cat/blog/matt-s-trout/mstpan-1/
10:48 pink_mist Padre is pretty much completely irrelevant; it's an IDE, not a web framework :P
10:51 nic If you've ever had to walk a distance in really deep snow, and thought you were never going to make it, Mojolicious gives you snow shoes
10:51 nic (not literally)
10:51 pink_mist the one thing that Mojolicious does better than everything else is its async support ... but you haveto specifically code for it to make use of it
10:52 mib_926az2 lol sounds like this is the place to be then Mojo, I want to make all me usable perl scripts into a web front end for all the users of my org to see and use rather than having to log into many different servers to run etc.
10:53 mib_926az2 am I barking up the right tree
10:53 nic YES
10:53 pink_mist you mean NO
10:54 mib_926az2 oops is that a conflict
10:54 nic One of my uses is letting users browse and rerun data feeds...  which themselves are also written in something that uses Mojolicious
10:54 mib_926az2 let me re-phrase
10:54 pink_mist nic: his question was if he was barking up the wrong tree :P he's not :P
10:54 mib_926az2 lol
10:54 nic pink_mist: read it again
10:55 pink_mist oh ... huh
10:55 mib_926az2 my phrase was a funny flip
10:55 nic :)
10:55 pink_mist the *right* tree? 0_o well ok then :P
10:55 pink_mist nic++
10:55 pink_mist pink_mist--
10:56 * pink_mist must have looked at the start of the sentence and just assumed it would end like it usually does :P
11:15 sue joined #mojo
11:36 oliver joined #mojo
11:40 panshin_ joined #mojo
11:43 panshin joined #mojo
11:45 panshin joined #mojo
11:51 binlei joined #mojo
11:53 kes joined #mojo
12:12 jberger mib_926az2: Mojo is also very clean and easy to use while still being able to grow to large usage
12:12 jberger The code base is very tidy and incredibly well tested
12:13 jberger Those things plus async are why you should pick mojo
12:14 nic yes, the ratio between lines-of-code and qty-of-tests is a major selling point
12:16 jberger Oh and the fact that the entire system is designed for consistency
12:17 mib_926az2 seems like mojo does not like perl 5.10 will not install
12:18 mib_926az2 on my mac
12:18 nic You need 5.10.1 or higher
12:18 n16gel joined #mojo
12:18 nic It's best to have 'without threads' but you probably won't care about that for a long while
12:19 mib_926az2 hmm not sure if my mac will run 5.10.1 never had to upgrade perl on a mac let me think
12:20 kaare joined #mojo
12:20 nic btw, v5.22.1 has big performance gains; only stick with 5.10 if you _really_ have to
12:21 nic mib_926az2: Are you using homebrew?
12:21 pink_mist mib_926az2: I heard homebrew or something like that will allow you to have a separate perl
12:22 nic best to leave the system perl alone
12:22 mib_926az2 ok so if I build my nice new flashy front end in mojo using the latest version of perl.  Will the end result be transferrable to a linux host running standard perl apache etc
12:23 nic for a lot of projects it's seamless
12:23 pink_mist as long as the perl it's running isn't old as heck =)
12:23 mib_926az2 or used I be developing this on the Linux server which is not cool
12:24 mib_926az2 This is perl, v5.8.8 built for x86_64-linux-thread-multi  Copyright 1987-2006, Larry Wall
12:25 pink_mist that's way too old
12:25 pink_mist a decade old even
12:25 nic For the most part the differences you'll encounter are if/when optimising hypnotoad for performance, and that installing stuff on mac is somewhat harder
12:25 mib_926az2 that's what the linux distro came with
12:25 nic What distro?
12:26 mib_926az2 OEL
12:26 nic poor you
12:26 nic Debian stable currently comes with v5.20.2
12:26 pink_mist mib_926az2: well no worries, you can use perlbrew to brew a newer perl for yourself =)
12:26 nic but I ignore that and install my own in /opt/perl
12:27 mib_926az2 I am sure this production box will have dependency on the old perl
12:27 mib_926az2 not sure I want to muck about with that
12:27 nic I already said, leave the system perl alone ;)
12:27 mib_926az2 perlbrew might be an option
12:28 nic My personal opinion is only a madman would use perlbrew on a production server
12:28 panshin joined #mojo
12:28 moritz nic: so, what do you use?
12:28 nic standard perl build
12:29 binlei joined #mojo
12:29 nic Each box has as many perl versions as it wants, but without the complexity of perlbrew
12:30 moritz so you do it manually?
12:30 mib_926az2 should I just install the latest verion for the distro in /opt/perl
12:30 nic ./Configure -des; make && make test && make install
12:30 nic mib_926az2: Do you need to worry about the production box at this stage?
12:30 mib_926az2 yes
12:31 nic There's a big gap 'tween what's good for learning (in dev) and what's good for production
12:31 moritz nic: that would be too much compiling and manual dependency management for me
12:31 mib_926az2 indeed
12:31 mib_926az2 CGI.pm ?
12:31 nic moritz: it's no more than you have with perlbrew
12:32 nic Doing that with v5.10 and earlier was quite a headache (so I didn't do it) but since v5.12 it's been a breeze
12:33 moritz nic: if you don't even add a --prefix to Configure, you'll overwrite the previous installation, no?
12:34 moritz nic: somehow I get the impression that there are more moving parts to the whole thing than is visible upfront
12:34 moritz nic: how do you install perl modules? do you bootstrap a cpanminus for each perl?
12:34 mib_926az2 no perl modules are installed just core perl on a linux distro
12:35 nic from https://cpanmin.us -- it's one line
12:36 moritz ah well, to each his own
12:38 mib_926az2 Unfortunately it is not a play ground a we have to work with the tools that production boxes come with, scripting is fine.
12:39 mib_926az2 What tool would be best to make a nice HTML front end served from a Linux box running perl 5.8 and most likely appache somewhere on it
12:39 nic I also install IO::Socket::SSL in amongst the core modules, just for convenience
12:40 moritz I'm pretty sure my ops team would ignore me if I'd instruct them to compile a perl on each server, bootstrap a cpanminus from the Internet and what not
12:40 mib_926az2 yes I agree
12:41 nic I don't  -- I give them a .deb that they install on their local repo
12:41 nic moritz: This is instead of perlbrew -- are you seriously saying your ops team would prefer perlbrew?
12:41 mib_926az2 when you are running thousands of servers you do not add no standard bundles to the mix very often
12:42 moritz nic: no; I'm saying that they need something simpler
12:42 nic mib_926az2: We're running almost one thousand servers
12:42 moritz nic: so, do you ship your perl binary in the .deb package?
12:42 nic yes
12:42 moritz ah, now we're getting closer
12:43 moritz so you basically fat-package
12:43 moritz is there a nice debhelper that does that?
12:43 moritz I'd love to have something like dh-virtualenv (python) for perl
12:43 nic also a .rpm since we shook hands with the dark side a couple of years ago
12:44 nic hmm, I haven't looked at that; that would help a lot of people
12:47 nic I've also wondered about a public deb repo for perls that install under /opt/perl
12:49 mib_926az2 Hi sorry to stray back to the original problem, are we saying now Mojo will be of no use to me ?
12:49 nic moritz: To answer one of your earlier questions, the magic sauce for the one-line build is having a Policy.sh file
12:49 nic mib_926az2: I didn't see anyone say that; I believe everyone said the opposite
12:50 nic It's fairly easy to install a chosen version of perl on a mac dev box
12:50 mib_926az2 true
12:50 nic and it's fairly easy to install a chosen version of perl on a linux production box
12:51 nic How the latter is done is up to the people maintaining the box
12:52 nic btw, I hate redhat, but I had to retract my blocker on our business migrating to it when I discovered building a custom perl was nearly as easy as on debian
12:52 moritz nic: so, another step necessary: copy the policy file :-)
12:52 nic only on the box doing the building
12:53 mib_926az2 sure, I support the Linux box so I can look at that.  If I develop using Mojo on a mac will it interact at all with the system perl version 5.8 on the Linux host or is that not going to work at all ?
12:54 nic I think we copy three files into a new src tree: cpanminus, Policy.sh, and .env (for making global module installs easy, even tho those are discouraged)
12:55 nic mib_926az2: there is a version of Mojolicious backported to v5.8, but I cannot imagine that being your best option
12:57 CandyAngel Has anyone used Mojo::(Reactor|IOLoop) as the heart of an interactive application (e.g. OpenGL)?
12:57 moritz sorry, but betting on perl 5.8 means losing
13:01 mib_926az2 I would be intersted in the backported version
13:01 mib_926az2 where is that ?
13:02 nic https://github.com/jamadam/mojo-legacy
13:06 sue joined #mojo
13:09 Grinnz mib_926az2, perl-build is the easiest way to build your own perl, no need to muck about with configuring
13:09 Grinnz https://metacpan.org/pod/distribution/Perl-Build/script/perl-build
13:09 Shaeto left #mojo
13:09 Grinnz mib_926az2, whatever perl you use to install Mojolicious is the one it will run under, like any well behaved cpan module
13:12 punter joined #mojo
13:13 nic "perl you use to install" is quite subtle and not clear to a lot of people
13:13 n16gel joined #mojo
13:20 ajr_ joined #mojo
13:22 jberger mib_926az2: building new code for an ancient perl and a back ported mojo :'(
13:26 jberger Mojo legacy is only mojo version v5.56
13:26 jberger jamadam kept that very up to date for so long
13:27 jberger I wonder if he's still using it at that version or if he finally got to upgrade his perl
13:42 neilhwatson joined #mojo
13:46 vicash hello. when writing unit tests for routes that involve a backend database such as Postgres, what is the best practice for testing ? do i create a database just for tests and skip those tests if the db is not available or accessible ?
13:47 neilhwatson I have initization tests that create a test database.
13:48 moritz that, or separate the database calls out into a model, and supply a mock model for testing
13:48 vicash neilhwatson: do you also drop the database at the end of the tests ?
13:48 neilhwatson yes
13:48 anparker joined #mojo
13:48 neilhwatson https://github.com/evolvethinking/delta_reporting/tree/master/app/t
13:48 neilhwatson see 10 and 999
13:49 moritz if one writes database-agnostic SQL, one can also use an in-memory sqlite DB for testing
13:49 Grinnz nic, indeed, it was a bit of a eureka moment for me
13:49 moritz still has to be initialized, but saves you the trouble of dropping it at the end
13:49 moritz or of several tests running at once interfering with each other
13:51 Grinnz vicash, see https://metacpan.org/pod/DBIx::TempDB
13:51 vicash moritz: interesting.
13:52 Grinnz i wouldn't bother trying to write database-agnostic SQL. there's too much variation and you'll lose the ability to optimize certain things. using modules with pluggable database backends is an option, though
13:53 vicash Grinnz: i am using Mojo::Pg so can DBIx::TempDB be used with it ?
13:53 Grinnz DBIx::TempDB is batman's solution for basically what neilhwatson is doing
13:53 Grinnz sure. Mojo::Pg just uses DBD::Pg, which DBIx::TempDB can also use
13:54 vicash ok thanks. i will play around with it to see how it works with Migrations.
13:56 Grinnz the "url" method can be used to get a connection URL that can be passed directly to Mojo::Pg to connect to the temporarily created database
13:56 vicash Grinnz: understood.
14:01 panshin joined #mojo
14:07 dod joined #mojo
14:14 orev joined #mojo
14:19 taichi joined #mojo
14:26 batman vicash: it supports almost any DBD::whatever, but there's explicit support for DBD::SQLite, ::mysql and ::Pg
14:27 batman vicash: there's one default i've forgotten to change, and that's new(drop_from_child => 1)
14:27 batman i would suggest using (drop_from_child => 1), since it's more robust
14:27 mcsnolte joined #mojo
14:28 * batman will change the default in the next release
14:30 zivester joined #mojo
15:06 dod joined #mojo
15:14 asarch joined #mojo
15:16 binary My Mojo::UserAgent is trying to connect via ipv6. It's getting an ipv6 address when doing a domain lookup. Wtf
15:16 jberger binary: wtf wtf?
15:16 mspo that's your os
15:17 binary Ah. Can you point me in the right direction, mspo?
15:17 binary I'm using ubuntu 14
15:18 mspo you need to configure /etc/gai.conf
15:18 mspo the format is really weird though so good luck ;)
15:18 mspo a quick google shows something like this might work: precedence ::ffff:0:0/96  100
15:19 binary Thanks!
15:19 mspo netbsd is adding a sysctl to change it too
15:19 mspo http://forevercached.syphzero.net/2012/09/preferring-ipv4-over-ipv6.html
15:19 binary Yeah I see it's a new thing to help IPv6 takeover. How annoying
15:20 mspo "new" since like 2010 but okay :)
15:20 jberger is there something wrong with ipv6? why does it matter?
15:20 mspo it's common that ipv6 will timeout for a lot of people
15:20 binary That's it sorted now :)
15:21 mspo for example your box has an ipv6 address but your wireless router doesn't support it
15:21 mspo or your rstold died or your firewall isn't setup for it or whatever
15:21 jberger ¯\_(ツ)_/¯
15:21 mspo so you wait for initial connections to the AAAA to timeout and then move on to the A
15:21 binary jberger: And because a lot of sites will only have an ipv4 website. Mine for example. My domains have an ipv6 address, but I don't have any ipv6 setup
15:22 jberger binary: that sounds like something you should fix on your domains
15:22 mspo yeah :)
15:22 binary :)
15:23 mspo if you're interested- gai.conf stands for GetAddrInfo
15:23 binary Useful to know
15:25 binary Ah, it's linode. I'm using them for dns and they automatically add some ipv6 address
15:25 binary I'll remove it
15:27 neilhwatson pesky ipv6, it'll never catch on.
15:27 mspo I actually have it (comcast) at home now
15:28 abra_ joined #mojo
15:28 mspo and mspo.com *can* have it but I was having rtsold issues and was just too lazy to figure them out
15:28 neilhwatson Only tunnels here in Canada, the backwoods of the Internet :(
15:30 Grinnz_ the mobile networks seem to be using ipv6 pretty well
15:31 berov left #mojo
15:35 bowtie joined #mojo
15:51 gryphon joined #mojo
15:51 kivilahtio joined #mojo
15:52 sue joined #mojo
15:53 n16gel joined #mojo
16:16 jontaylor joined #mojo
16:17 lluad joined #mojo
16:18 sh4 joined #mojo
16:28 n16gel joined #mojo
16:29 zivester joined #mojo
16:48 hesh joined #mojo
16:51 hesh Hi all... I have a project requiring both a cli and web interface. is mojolicious a good fit for that scenario?
16:52 lluad It likely is, yes.
16:52 jberger mojolicious application have a commands system built in
16:52 jberger hesh: what do you envision doing with the commands?
16:53 jberger (e.g. what do you envision doing with the cli)
17:03 neilhwatson I do this, it is not hard.
17:13 sri last chance to comment on this before i close it! https://github.com/kraih/minion/pull/22
17:16 jberger sri: I think once 9.5 is required it might be kinda cool
17:18 hesh the project is generally for document searching. want to use cli for controlling indexing and storage while using web interface for searching
17:18 jberger hesh: I did exactly that at $job[-2]
17:19 preaction i think avkhozov is trying to say that if you have this lock option, you can make cron trigger minion tasks without worrying about duplication. you could also have multiple servers triggering minion jobs via cron, for redundancy. only one would win
17:19 jberger hesh: had cli commands that would do things like reindex elasticsearch
17:19 preaction i think it needs to be explained better in the docs what it could be used for, that might improve its desirability
17:19 jberger and ES was used in the web interface for searching
17:20 sri no other job queue has unique jobs implemented like that
17:20 sri at least to my knowledge
17:21 sri "it sounds kinda cool" doesn't really qualify it for inclusion imo
17:22 jberger no, and I don't have a use for it personally, but it has come up before
17:22 jberger I never did get my answer on performance impact for those not using it
17:22 sri and when it came up before, the person requesting it specifically didn't want the "lock" solution :)
17:23 jberger how does unique=>1 work?
17:23 jberger does it hash all the arguments or something?
17:23 sri nope
17:24 sri you just can't insert more jobs for a task if there is one in state active/inactive with unique flag
17:24 jberger IIRC back the last time it came up I fell on the side of this kind of solution too, that way you can implement locking as you need it, all you get is the primative
17:25 jberger so that would be a unique task rather than a unique job
17:25 sri the last discussion btw. https://github.com/kraih/minion/issues/7
17:25 jberger \o/ I am consistent at least!
17:25 hesh jberger: i assume you were happy with your mojo cli/web sol'n
17:26 jberger hesh: yeah, it worked great
17:26 jberger abstract most of the hard stuff into business-logic (model?) classes and then just expose them to the web and cli in the appropriate ways
17:26 jberger but it worked like a charm
17:27 hesh jberger: yep. that's the plan. thanx. i'll move ahead with mojo
17:28 sri jberger: actually, i think what all those people really want is not unique jobs, but periodic jobs https://github.com/mperham/sidekiq/wiki/Ent-Periodic-Jobs
17:28 jberger that would be a seriously cool feature
17:29 jberger I had that problem at $job[-2] where our cron host was subject to failover and knowing how to fail cron to another box was ... over my head
17:31 dod joined #mojo
17:34 PryMar56 joined #mojo
17:34 jberger sri: my stickers shipped
17:34 sri \o/
17:38 sri i guess periodic jobs could actually could use the lock => key solution from #22
17:38 sri s/could//
17:38 dod joined #mojo
17:40 sri you make a recurring job timing table like app->minion->recurring('2 * * * * *', 'foo') at startup next to your ->add_task() calls, and then every worker checks the table every minute while it is running for jobs that need to be enqueued
17:40 sri and those jobs use a lock
17:53 dim2 joined #mojo
18:11 c--_ joined #mojo
18:14 binary Why can't I create non-blocking Mojo::UserAgent jobs inside the callback of a Mojo::IOLoop->timer?
18:18 binary Or an EV::timer
18:18 binary Basically it seems I can't create any ua's once the event loop begins
18:27 n16gel joined #mojo
18:38 ajr_ joined #mojo
18:39 Grinnz_ you can; non-blocking Mojo::UA calls use the singleton loop
18:40 sri looks like binary is cross posting in multiple channels
18:44 sri binary's question is actually very easy to answer
18:45 sri but now we let #perl figure it out ;p
18:46 preaction i'm the poor sucker answering in #perl and i'm stumped
18:46 disputin joined #mojo
18:47 sri one might say it's a frequently asked question
18:47 Grinnz_ i figured it out, its scoping
18:47 sri STOP!
18:48 preaction tm604 got it before me, and without help :(
18:48 Grinnz_ the FAQ doesn't really say the cause
18:48 sri now i feel like i rewarded cross posting :(
18:48 preaction Grinnz_: "the user agent got destroyed, which forces all connections to be closed immediately."
18:48 Grinnz_ hmm, i guess
18:49 sri binary: just to be clear, cross posting is pretty rude
18:51 Grinnz_ he didn't get a response here for 10 minutes before posting there, i think thats reasonable
18:56 binary sri: Oh, I didn't realize. My apologies. It's sorted now anyway!
18:57 * binary reads up and laughs
19:12 c--__ joined #mojo
19:15 n16gel joined #mojo
19:45 panshin joined #mojo
20:03 n16gel joined #mojo
20:14 oliver hah, I'll be implementing scheduled jobs with minion soonish, but was planning to do it the same way I've done before (with Pg <9.4 backed queue) ... allow dupes to be queued but delete duplicates immediately after dequeue (all in a transaction). Using Algorithm::Cron for the scheduling.
20:15 oliver if something better comes along then great, but this seems to work well enough for me.
20:16 sri don't think anyone is working on a solution for scheduled/periodic jobs yet
20:17 oliver sri: ok
20:19 mspo http://search.cpan.org/~cornelius/AnyEvent-Cron-0.03/lib/AnyEvent/Cron.pm ?
20:19 sri doesn't solve the problem of duplicated jobs
20:21 sri when you schedule a job every 5 minutes, you want only one copy of it to run at a time
20:21 mspo flock it
20:21 sri you have to consider a backlog of maybe thousands of jobs
20:21 sri you have a distributed worker queue with many machines
20:22 sri s/queue/pool/
20:24 sri oh, you also have to condier retries, you don't want your scheduler to queue a new job while the old one is still being retried
20:24 sri *consider
20:24 mspo I do actually deal with this
20:25 mspo with a big laravel/artisan shitshow
20:25 mspo of thousands of jobs :)
20:25 mspo but in the simplest case we lock to redis or flock on the local filesystem; you need a job-by-job cofig
20:25 sri funny thing is, this version of shared lock can actually deal with most of the problems https://github.com/kraih/minion/pull/22
20:26 sri so it's not a bad primitive
20:26 mspo if you configure a job to run every five minutes you have to decide what to do if it takes seven minutes to run
20:26 mspo start the next one immediately, wait for the nex timeslice, interrupt the still-running job, go ahead and execute on another node
20:27 sri i don't care too much abot custom solutions
20:28 sri more interested in established solutions shipping with popular job queues
20:28 mspo there's also this: http://nerds.airbnb.com/introducing-chronos/
20:28 mspo https://laravel.com/docs/5.1/queues
20:28 sri think this might be the most commonly used one https://github.com/mperham/sidekiq/wiki/Ent-Periodic-Jobs
20:29 sri mspo: i don't see anythign about periodic jobs there
20:30 mspo of course most java systems use quartz
20:30 mspo sorry it's the same commands but a different option: https://laravel.com/docs/5.1/scheduling
20:32 sri ah, it uses one server with a cron entry for scheduling
20:32 mspo yeah laravel isn't doing distributed stuff
20:33 mspo I think rails has something similar but I forget the name
20:33 mspo it installs crons
20:33 go|dfish joined #mojo
20:46 sri well, i've opened an issue for discussion https://github.com/kraih/minion/issues/24
20:46 sri collect information there
20:54 panshin_ joined #mojo
21:02 sri looks like postgres 9.5 support in travis is not a big priority... they are slacking for two weeks now https://github.com/travis-ci/travis-ci/issues/4264
21:02 genio It took them quite some time to get Perl 5.22 as well, though
21:20 ajr_ joined #mojo
21:29 sri hmm, even manually i can't get postgres 9.5 working on travis
21:39 sri re unique jobs, it is funny how not using sort in Mojo::JSON now makes this feature harder in Minion
21:40 sri since you can't just use the generated json to make a checksum
21:50 sri you'd think Grinnz_ and bpmedley would have an opinion on locking in minion too
22:03 jberger I think I'm going to eventually take bpmedley's opinions on that
22:04 sri jberger: btw. regarding performance, i think it's irrelevant
22:04 sri since we'd just use a unique constraint on the field
22:04 jberger it probably is, but it was a valid question and it got ignored
22:04 jberger :s
22:04 sri yea
22:05 sri avkhozov really doesn't do a good job selling his features
22:05 jberger he is enthusiastic, which is worth something
22:06 sri yea, and he's often right too
22:07 sri think language barrier is a big issue
22:07 sri really too bad
22:08 sri somehow i have a feeling #22 will actually make it into minion 5.0 in some form, and scheduling will not ;p
22:19 disputin joined #mojo
22:47 Kogurr joined #mojo
23:19 odc joined #mojo

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