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

IRC log for #mojo, 2016-08-08

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

All times shown according to UTC.

Time Nick Message
00:36 deserted joined #mojo
02:25 noganex_ joined #mojo
02:59 aborazmeh joined #mojo
03:09 lluad joined #mojo
03:21 sri batman: that's easy, modifications of those structures are unsupported
03:35 sri interesting how there seems to be a scalability wall around 90k rps https://github.com/kraih/mojo/issues/925#issuecomment-238066683
03:37 sri maybe i was wrong and there is something else than uneven load balancing at play
03:52 lluad joined #mojo
04:38 sri Grinnz: overwatch is definitely fun, not sure it's enough content to justify the price though
04:41 Grinnz any heroes or maps they add will be free though
04:44 dave I believe the "content" is your desire to compete and get rank
04:45 dave and you balance that with your lack of desire to deal with the average player on battle.net ;)
04:47 sri yea, the diversity of the heroes definitely makes the game
04:48 dave you should put together an all mojo core dev team
04:59 inokenty-w joined #mojo
05:07 mpapec sri: yell if you have new ideas for benchmark
05:07 sri mpapec: out of ideas for now, even though i'm curious what's causing the 90k rps wall
05:08 mpapec did you make graphs which show that?
05:09 sri mpyou have 72 physical cores right?
05:10 sri s/mp//
05:10 mpapec HT is enabled, so half of that!
05:10 sri oooh!
05:10 sri that changes everything i guess :)
05:11 sri ok, now it all makes sense
05:11 mpapec that's why I asked you about HT :)
05:11 sri my 20% figure was for small numbers of cpu cores, i actually have no idea how it works with that many
05:13 mpapec so, when calculating how many workers to employ for hypno
05:13 mpapec it only makes sense to take real cores in consideration
05:15 sri yea, i suppose so
05:15 mpapec btw, did EV show any improvements?
05:17 sri i've not drawn curves, but EV will be better the more connections you have
05:17 sri per worker
05:18 mpapec for given bmark that means first half of the curve
05:19 sri from what i can see the numbers make perfect sense
05:19 mpapec great
05:19 sri mpapec++
05:21 sri mpapec: did you configure EV?
05:21 mpapec nope!
05:21 sri by default it uses select()
05:21 mpapec just used it
05:21 mpapec ah
05:21 Grinnz epoll on linux in my experience
05:22 sri by default?
05:22 Grinnz yea
05:22 mpapec how to make it show what it uses?
05:22 sri interesting, on BSD it avoids kqueue and defaults to select
05:22 Grinnz my tests for POE::Loop::EV print out the default cause its weird
05:23 Grinnz https://metacpan.org/source/DBOOK/POE-Loop-EV-0.13/t/00_info.t
05:23 sri i use dtruss on os x to watch the syscalls the process makes
05:25 mpapec perl -MEV -E 'say EV::backend'
05:25 mpapec 4
05:25 sri that's epoll
05:25 sri http://mojolicious.org/perldoc/Mojo/IOLoop#DESCRIPTION
05:27 sri jberger: i guess that means in the next framework benchmarks we'll be higher up ;p
05:28 mpapec yes, epoll
05:28 mpapec perl -MEV -E'
05:28 mpapec say +{
05:28 mpapec EV::BACKEND_SELECT()  => "select",
05:28 mpapec EV::BACKEND_POLL()    => "poll",
05:28 mpapec EV::BACKEND_EPOLL()   => "epoll",
05:28 sri no!
05:28 mpapec EV::BACKEND_KQUEUE()  => "kqueue",
05:29 mpapec EV::BACKEND_DEVPOLL() => "devpoll",
05:29 mpapec EV::BACKEND_PORT()    => "port",
05:29 mpapec }->{EV::backend()}
05:29 mpapec '
05:29 sri please use a paste site next time
05:30 mpapec ah, yes
05:31 mpapec which mechanism make more sense on linux?
05:31 sri epoll is correct
05:31 mpapec are there any epoll fine tuning settings?
05:38 disputin joined #mojo
06:03 VVelox joined #mojo
06:41 Vandal joined #mojo
07:15 batman sri: hehe, ok :-) but still... i find it curious, since i thought that object was local to the current $controller
08:07 osfabibisi joined #mojo
08:09 * batman has run out of vacations now
08:16 trone joined #mojo
08:17 Jonis :/
08:17 Jonis batman: I just came back to work from mine today.
08:18 Jonis from 25-30 degrees every day in Marseille, to 13 degrees and the floodgates opening in Bergen
08:18 Jonis does not compute
08:18 purl DANGER!  DANGER WILL ROBINSON!
08:19 Jonis <3 purl
08:22 ashimema_ joined #mojo
08:31 sri batman: oh, my bad, misunderstood
08:32 sri mixed it up with the earlier route pattern thing
08:32 vytas joined #mojo
08:32 sri no idea what's going on there
08:39 abra joined #mojo
08:40 Atog joined #mojo
08:56 osfabibisi joined #mojo
09:45 kes joined #mojo
09:57 meshl joined #mojo
10:09 batman Jonis: hehe. yeah... it's raining like crazy in oslo as well :(
10:10 batman sri: no... I find it very curious... I will try to look into it later.
10:20 mpapec heh, http://www.zerohedge.com/news/2016-08-07/cnn-host-slams-americas-greatest-olympian-ever-not-being-black-muslim-woman
10:20 itaipu joined #mojo
10:30 kid51 joined #mojo
10:36 kaare joined #mojo
10:41 sed joined #mojo
10:42 sed Hi all,
10:42 purl it has been said that Hi all, is it possible to do http://wiki.catalystframework.org/wiki/wikicookbook/urlpathprefixing, but maybe in a more "standard" way using Root.pm (to catch the first level of the request URL) ?
10:43 sed I have the error when rendering template
10:43 sed Can't locate object method "_bundled" via package "Mojolicious::Renderer" at template not_found.sso_staging.html.ep line 77.
10:43 sed 72:         margin: 0 auto; 73:       } 74:     % end 75:   </head> 76:   <body onload="prettyPrint()"> 77:     %= include inline => app->renderer->_bundled('mojobar')
10:43 sed i have mojo 6.66
10:44 sed so app->renderer->_bundled('mojobar') is not in 6.66 i guess
10:44 sed what options is available?
10:48 inokenty joined #mojo
11:02 tchaves joined #mojo
11:07 Kripton joined #mojo
11:12 tchaves joined #mojo
11:25 Kripton joined #mojo
11:28 dotan sed: looks like you have your own template (not_found.sso_staging.html.ep) containing copy-pasted code from an old version of Mojolicious' built-in error pages.
11:30 dotan Probably you could replace the outdated bits with stuff from the 6.66 built-in templates? https://github.com/kraih/mojo/tree/v6.66/lib/Mojolicious/resources/templates/mojo
11:55 itaipu joined #mojo
12:04 jberger purl: forget Hi all,
12:04 purl jberger: I forgot hi all,
12:08 jberger Well it seems that sed left but for the record _bundled is/was a private method and isn't for use
12:08 jberger As are the bundled static files
12:08 izz joined #mojo
12:12 izz .
12:14 mpapec anyone going? https://www.perl.dance/
12:18 kid51 joined #mojo
12:18 mpapec http://act.yapc.eu/alpineperl2016/talks :o
12:19 mpapec (another conference)
12:21 * Lee is going to APW
12:37 gizmomathboy joined #mojo
12:41 meshl joined #mojo
12:50 neilhwatson joined #mojo
12:52 jberger mpapec: there isn't a large overlap between mojo and dancer users in my experience
12:55 Janos joined #mojo
13:04 izz joined #mojo
13:04 asarch joined #mojo
13:25 jnbek joined #mojo
13:33 orev joined #mojo
13:58 mcsnolte joined #mojo
14:18 itaipu joined #mojo
14:29 ptolemarch joined #mojo
14:53 dotan "If you get a rainbow unicorn t-rex error after upgrading, you might be missing a new CPAN dependency"
14:53 dotan You have to admire Yuki Kimoto's attempt at making an installable Mojolicious app
14:54 dotan https://github.com/yuki-kimoto/gitprep
14:56 Grinnz hmm... yet it's not on cpan... and not using Mojo::SQLite :P
14:59 dotan His project is much older than Mojo::SQLite
15:16 jberger dotan: ?
15:16 jberger there are already installable mojolicious apps
15:21 lluad joined #mojo
15:45 dotan jberger: I'm not saying his project's unique, I'm just admiring the attention to nitty-gritty details.
15:46 jberger ah, that's not how I read it
15:50 cuechan joined #mojo
15:57 dotan Sorry. It wasn't a dig at Mojolicious - the installation difficulties would be common to any Perl web app.
15:58 jberger of the things I'm proud about for Galileo, which itself isn't the most versatile of apps, the installation process is pretty nice
15:58 dotan Although gitprep demonstrates the little-publicized fact that many mojolicious apps can run as plain CGI on shared hosting
16:00 jberger with a little care, and if you don't need websockets, all should be able to
16:00 jberger then again, mostly my apps do use websockets
16:00 jberger so ... I guess my statement isn't that true
16:00 jberger :-P
16:02 dotan jberger: you use websockets because it's there, it's cool and it's relatively unique to Mojolicious. But there's no real reason for Galileo to use them, is there?
16:02 jberger no there isn't
16:03 jberger the roadmap for a Galileo successor would remove that requirement and use XHR (ajax) for most things
16:03 jberger then again, it would be able to use websockets for something a little more useful (if I can ever make it work)
16:12 dotan websockets might be useful for a CMS if it had multiple users working closely together
16:17 jberger yes, that's the idea
16:18 jberger I would like to have a CMS with collaboratively editable documents
16:18 jberger the problem so far is finding a relatively language agnostic implementation of the various algorithms for collaborative editing
16:21 dotan https://twitter.com/Beschizza/status/761834300132225024/photo/1
16:21 dotan That would be very cool
16:27 disputin joined #mojo
16:29 jberger for static pages we already have Statocles, if you haven't looked at that yet
16:34 lluad What I'd love for a blog would be static pages + dynamic comments. I've not found anything that does comments that doesn't suck a lot more than wordpress, though.
16:39 Grinnz_ what, you don't like discourse?!
16:40 lluad Is discourse embeddable?
16:40 lluad Maybe I should take another look at it.
16:41 Grinnz_ that was sarcasm, nobody likes discourse :P
16:42 lluad Oh. Good.
16:42 lluad I'd previously skipped over it as horrible. :)
16:43 lluad . o O ( too early in the day for lluad to recognize any sarcasm other than the most brutally obvious )
16:44 preaction lluad: i use Disqus with my Statocles site, but I also know facebook has a way to do it (that i have not yet integrated with statocles), and i've stalled on a custom app for it (Dynamocles)
16:44 lluad Yeah, I ended up at Disqus as "this is the best I found" and it still wasn't as good as wordpress, with all it's warts.
16:44 Bean joined #mojo
16:45 lluad Some quiet weekend I should just write the thing. How hard can it be? :)
16:50 dotan jberger: I rooted around the Statocles site the other day, curious how it handled YAML front matter. My browsing went: https://pgexercises.com/ -> codemirror -> YAML frontmatter -> Jekyll docs -> Statocles
16:50 dotan https://pgexercises.com/ is cool, BTW
16:52 preaction dotan: this is how it does it, https://github.com/preaction/Statocles/blob/master/lib/Statocles/Store.pm#L202
16:53 dotan lluad: I remember when blogging was becoming a thing and comment spam started popping up and someone blogged that bloggers are reinventing all the problems of email...
16:53 lluad dotan: Very much so.
16:53 preaction sure, but everyone reinvents that. like slack
16:54 lluad Comment spam is very, very naive compared to email spam, though. Akismet pretty much kills it dead.
16:55 preaction lluad: as for "how hard can it be?" the issue becomes authentication/authorization and security. everything else is pretty trivial.
16:56 dotan preaction: R
16:56 dotan I meant Right, or thanks, sorry.
16:56 preaction and that's basically where i stopped. authing the user, authing the site (CORS? reverse proxy?), and then doing the right thing
17:02 sri haha, so i'm not the only one who hates discourse :)
17:15 Grinnz_ disqus is rather annoying too (as a user)
17:16 Grinnz_ anything where you have to sign in to something that's not-the-site just feels wrong
17:16 Grinnz_ for comments, specifically
17:27 jberger Grinnz_: you are going to need something like that though if you want comments on static blogs
17:27 jberger the debate over the value of static blogs aside
17:27 Grinnz_ i suppose
17:29 jberger Sinan Unur just puts a link to the reddit discussion for each blog post at the bottom of the post
17:29 jberger that kinda works too
17:33 PryMar56 joined #mojo
17:33 Grinnz_ mm, reddit is a little more palatable
17:34 preaction i looked into auto-posting the blogs to reddit, but didn't take that step
17:34 preaction the issue of idemopotency comes in to play: it should never, ever be posted twice
17:44 disputin joined #mojo
18:15 TheGrinnz joined #mojo
18:16 itaipu joined #mojo
18:34 trone joined #mojo
18:40 zivester joined #mojo
18:57 itaipu joined #mojo
19:36 disputin joined #mojo
19:59 jberger so it seems that there is a template app for Dancer2 on openshift: https://www.openshift.com/container-platform/features.html
19:59 jberger if someone would like to contribute a Mojo one that would be a cool way for someone to contribute!
20:00 sri how odd, i remember a mojo example for openshift
20:23 jabberwok joined #mojo
20:56 kid51 joined #mojo
21:10 disputin joined #mojo
21:27 Zx3 joined #mojo
22:20 Janos joined #mojo
23:33 disputin joined #mojo

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