Camelia, the Perl 6 bug

IRC log for #mojo, 2010-07-23

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

All times shown according to UTC.

Time Nick Message
00:59 arthas joined #mojo
01:18 vel joined #mojo
01:50 doubi joined #mojo
02:21 ask joined #mojo
02:45 Alias_ joined #mojo
03:19 su-bzero_ joined #mojo
03:20 Alias_ joined #mojo
03:32 janus joined #mojo
03:59 ashleydev joined #mojo
04:04 tl joined #mojo
04:54 kimoto_ joined #mojo
06:04 yakudzo joined #mojo
06:12 yko joined #mojo
06:18 dotan joined #mojo
06:18 ask joined #mojo
06:24 baton8 joined #mojo
06:44 ask joined #mojo
07:02 Alias_ joined #mojo
07:43 esskar joined #mojo
07:47 esskar moin
07:48 zakame hi
09:08 sri joined #mojo
09:43 kimoto joined #mojo
09:59 * sri yawns
11:01 Alias_ joined #mojo
11:04 vti sri: when using daemon_prefork i can't share some variables of course.. but using daemon is not efficient, i am thinking about building a daemon backend that controls all the shells and a frontend daemon_prefork that connects via websockets to backend and transfers all the information from backend to the browser, is this insane?
11:05 sri vti: why would you need to share information between processes?
11:06 vti because i want several users to see the same shell
11:07 sri ah
11:07 sri sounds ok then
11:08 vti can i open a websocket connection inside a websocket connection? :)
11:09 sri of course, why wouldn't you?
11:10 sri but i think it might be a better idea to not proxy websocket connections and just keep a pool of normal daemon
11:10 sri *+s
11:10 sri maybe on different ports
11:11 sri your app then just routes all shared connections to the same port
11:12 sri but i guess the details don't matter yet
11:14 vti maybe since most of the connections are websockets anyway, daemon at frontend is not a bad idea
11:30 vti if you google 'daemon prefork' mojo is at the first results :D
11:30 sri :D
11:32 sri wonder how the rails folks are planning to implement a php like <% flush %>
11:32 sri http://php.net/manual/en/function.flush.php
11:32 garfield [ PHP: flush - Manual ]
11:32 sri as far as i can see it's pretty much impossible to get right
11:36 sri since they are rack based they'll have to return an iterator early, which would look exactly like the mojo version i posted to the list
11:37 sri and it still woudln't work in most deployment environments
11:48 sri http://github.com/oggy/template_streaming # yea, thats exactly how they've done it
11:48 garfield [ oggy's template_streaming at master - GitHub ]
11:48 jawnsy joined #mojo
11:49 sri they add a concept of prelayouts, that get rendered before the actual template
11:50 sri guess that would work
11:51 sri actually you don't even need flush
11:51 sri ->render(prelayout => 'foo') would be enough
11:52 sri it could automatically set a body callback splitting rendering in two parts
12:13 cognominal joined #mojo
12:36 esskar joined #mojo
12:37 esskar sri: http://perlpunks.de/paste/show/4c498cf6.e45.1f1 current github version
12:38 sri esskar: can't replicate, feel free to fix
12:40 sri must be a windows thing
12:41 sri if you enable debug messages it might be obvious
12:59 zakame exact same thing I reported on github
13:15 esskar memowe: http://perlpunks.de/paste/show/4c4995de.5403.273
13:16 esskar same happens with the version on CPAN
13:33 sri zakame: well, without a proper report i can't do anything :(
13:34 yko joined #mojo
13:35 dotan joined #mojo
13:48 sri zakame, esskar: how about one of you actually gives me the debug output?
13:50 sri as in MOJO_DEBUG_LEVEL=debug ./script/mojo test t/mojolicious/app.t
14:11 sri yay, it's raining finally!
14:11 sri
14:12 ashleydev joined #mojo
14:16 esskar sri: i dont see more output as i posted
14:16 sri impossible
14:16 sri ops
14:17 sri MOJO_LOG_LEVEL=debug
14:17 sri wrong env var
14:17 esskar oh, ok
14:20 esskar http://nopaste.voric.com/paste.php?f=d9wulc
14:20 garfield [ Nopaste ]
14:23 sri ok, that might answer it
14:25 sri something might be borked with handler detection on windows
14:26 sri not easy to debug without windows
14:29 sri anyone on windows got time to debug?
14:29 sri it's most likely in MojoX::Renderer around line 281
14:30 sri it might not be picking up the right template templates/foo/bar.rss.ep
14:33 sri there is nothing obvious broken, so i'll need someone with windows for debugging
14:35 sri (if there is interest in windows support)
14:35 sri i'll just move on for now
14:45 esskar sri: i will check it later
14:45 sri k
14:45 esskar rather busy
14:45 esskar got hacked!
14:45 sri :o
14:46 sri should be relatively simple, most likely just \ vs / again or so
14:46 sri oooh, i might have found it ;p
14:50 GitHub165 joined #mojo
14:50 GitHub165 mojo: master Sebastian Riedel * da6cf67 (1 files in 1 dirs): fixed handler detection for template files on windows - http://bit.ly/byVW38
14:50 GitHub165 left #mojo
14:50 sri esskar: please test :)
14:51 sri i think so far we only detected templates without a directory, thats why it wasn't discovered before
14:53 esskar sri: Result: PASS
14:53 sri \o/
14:53 sri yay for blind bug fixing
14:57 vel joined #mojo
14:59 yko wow
15:36 GitHub57 joined #mojo
15:36 GitHub57 mojo: master Sebastian Riedel * 5dc7bdd (1 files in 1 dirs): simplify example - http://bit.ly/9Ri1YG
15:36 GitHub57 left #mojo
15:41 ashleydev joined #mojo
16:22 sekimura joined #mojo
16:52 murata joined #mojo
17:11 ashleydev joined #mojo
17:34 stephen joined #mojo
18:08 GitHub50 joined #mojo
18:08 GitHub50 mojo: master Sebastian Riedel * 3d4f6b1 (3 files in 3 dirs): allow embedded apps to inherit params - http://bit.ly/aBOkW4
18:08 GitHub50 left #mojo
18:16 ask joined #mojo
18:23 xantus_ joined #mojo
18:24 ask joined #mojo
18:51 xantus_ Argument "6.55_02" isn't numeric in numeric lt (<) at Makefile.PL line 18.
18:52 xantus_ the make maker version
18:54 ask joined #mojo
18:54 xantus_ my ( $mm ) = $ExtUtils::MakeMaker::VERSION =~ /([^_]+)/; # that would fix it
18:55 xantus_ <3 mojo
18:55 sri and looks like a smiley
18:55 * sri likes it
18:56 xantus_ <3 looks like a triangle with a butt
18:56 xantus_ lol
18:56 sri lol
19:03 stephan48 sri, is it normal with the i18n plugin to get a blank page when no entry is found in dict?
19:03 sri huh?
19:03 stephan48 no matching entry for a i18n locup
19:04 sri is there nothing else on the page?
19:05 sri blank page is a very bad error description
19:06 xantus_ "Thanks to denial, I'm immortal"
19:07 stephan48 http://hpnew.stejau.de/?lang=en
19:07 stephan48 look at the test123 in the top left corner
19:07 sri i don't have time to play a guessing game...
19:07 garfield [ Stephan Jauernick> - Main Page ]
19:08 stephan48 http://hpnew.stejau.de/?lang=de with de its blank...
19:09 sri clicking that link it takes forever and i get nothing
19:10 stephan48 yea here too somehow the mojo app seems to block
19:10 sri unless thats intended, your app is broken
19:10 stephan48 mh....
19:12 xantus_ yeah, can't connect from here either
19:12 stephan48 http://abc123.pastebin.com/LGYdBQ4B <- thats my app nothing i think which could break in there
19:12 garfield [ Perl | #!/usr/bin/perl use strict; - Stephan48 - LGYdBQ4B - Pastebin.com ]
19:14 ashleydev joined #mojo
19:15 xantus_ sri: can we have something like __PACKAGE__->attr(json_class => $ENV{MOJO_JSON_CLASS} || 'Mojo::JSON');
19:15 sri xantus_: sure
19:15 xantus_ I would then just make a begin block check and set the env
19:16 xantus_ or perhaps a app->json_class
19:17 xantus_ lol, I see you added ojo.pm
19:17 xantus_ ice
19:17 xantus_ nice
19:18 xantus_ hah, I love it
19:19 stephan48 somewhen someone kills all these shortcut creates because he wants to use such a shortcut for usefull module :)
19:19 stephan48 shortcut creators*
19:20 xantus_ what
19:20 xantus_ you're very hard to understand sometimes
19:20 stephan48 i know.
19:20 stephan48 xantus_: atleast be happy that i dont talk to you in real life :)
19:20 xantus_ its ok :)
19:20 stephan48 then it would he hard
19:20 xantus_ lol
19:21 * vti drinks
19:21 xantus_ what are you drinking
19:22 vti today is the day!
19:22 xantus_ of what
19:22 vti i am too old after this day =/
19:22 xantus_ vti: happy birthday!
19:22 vti xantus_: yay!
19:22 stephan48 xantus_: today is his birthday :)
19:23 xantus_ Viva la Viacheslav!
19:23 vti the greatest!
19:23 vti :D
19:35 GitHub81 joined #mojo
19:35 GitHub81 mojo: master Sebastian Riedel * 277c3dd (3 files in 3 dirs): fixed pass through bug in Mojolicious::Plugin::I18n - http://bit.ly/cwC5Jg
19:35 GitHub81 left #mojo
19:35 sri stephan48: was a bug
19:36 stephan48 ....
19:36 stephan48 :)
19:36 stephan48 thanks sri
19:36 stephan48 where can i find the docs for uri_for?
19:37 sri some controller base class
19:38 stephan48 found it
19:39 stephan48 Mojolicious::Controller was it
19:42 spleenjack joined #mojo
19:51 cognominal joined #mojo
19:59 stephan48 the ep template system is awesome :)
20:00 stephan48 allowing me things like: <%= ($title ? ' - '.l($title) : '') %> or <%= link_to index => {%><%= l('Main Page') %><%}%>
20:00 sri it's just perl
20:01 stephan48 :)
20:02 und3f FCGI::PreFork doesn't work with lastes mojo
20:03 und3f how to run mojo with fcgi? :)
20:03 sri what changed?
20:04 sri it's not supposed to be broken, but it us unmaintained now i think
20:04 sri *is
20:04 und3f Can't locate object method "accept_lock" via package "Mojo::Server::FCGI::Prefork" at /usr/lib/perl5/site_perl/5.10.​1/Mojo/Server/FCGI/Prefork.pm line 29.
20:05 und3f got just this messages
20:05 und3f btw, were it is now? I am missed it new repo?
20:05 und3f or it is nowhere?
20:06 sri maintainer deleted the repo
20:06 und3f :|
20:06 sri it's also gone from cpan
20:06 und3f is other alternative for fastcgi?
20:06 sri psgi
20:07 sri or the normal fastcgi module
20:07 sri but i guess you want an external proc manager
20:08 und3f i need mojo's proc manager for nginx
20:08 vti new maintainer found! :)
20:08 vti hm
20:08 sri yea, you can have it if you want
20:08 und3f vti, there is nothing to maintain
20:08 vti sri: i thought it was und3f :)
20:08 sri i don't remember who it was :S
20:09 * und3f deleting rm -rf mojo-fcgi from $HISTFILE
20:09 und3f no, not me
20:09 vti hm
20:09 sri you can use psgi/plack or adopt the module :)
20:10 und3f sri, also i can understand how to run mojo with 'fastcgi' command
20:10 vti und3f: adopt!
20:10 und3f it just returns nothing
20:11 und3f it doesn't have proc manager?
20:11 sri und3f: it *is* a fastcgi script
20:11 sri nope, expects external proc manager
20:11 und3f got it now...
20:11 und3f thanks
20:12 sri it works just like a vanilla fastcgi
20:12 sri i think i heard that nginx has an external proc manager to run normal scripts
20:13 sri guess we'll let fcgi die then
20:13 fod joined #mojo
20:13 sri psgi should be enough to cover that area
20:15 und3f seems like people using spawn-fcgi
20:16 sri you could also switch to the normal prefork daemon and reverse proxy http
20:16 sri i like that option most personally
20:17 sri (but i'm a http nerd)
20:18 und3f I saw some tests about mojo in daemon mode and in fcgi with nginx, second is faster
20:18 sri i doubt you would notice a difference in a real world app
20:20 und3f i have mojo app that serves around 11k requests daily.
20:20 und3f could make some tests
20:20 sri thats 7 requests per minute ;p
20:21 sri i could write those http responses myself xD
20:21 yko or 1k requests in one minute and 6 per minute then
20:21 yko :)
20:22 * yko hides from whip.
20:22 sri seriously, as soon as you hit some backend service like a database webserver performance becomes irrelevant
20:23 und3f is template rendering lighter procedure than selecting? :)
20:23 sri selecting?
20:24 und3f from database
20:24 sri much
20:24 sri (depending on the template system though)
20:25 und3f MojoX::Renderer of course
20:25 sri those are compiled to perl code and cached in memory, so rather fast
20:29 sri the mojolicious daemon is not optimized for hello world examples, but more than enough for real world apps
20:31 pid joined #mojo
20:33 sri yko: for a mid sized app i get 500-900 req/s on my laptop
20:34 * yko still hides under the table.
20:36 yko sri: what kind of app?
20:36 yko Oh, forget it.
20:37 sri a test example with 40 routes, templates and layouts
20:37 sri stacked routes, no tree
20:37 yko wow
20:37 yko nice, ye
20:38 yko actually i don't care a lot about performance too
20:38 yko 4s/request in CGI mode it's also nice
20:38 sri i care a little about performance, but not micro benchmarks
20:39 yko Well.. actually i do care
20:39 sri the big question though is if it is fast enough
20:39 yko i have little Mojo::Lighter project - fat-free Mojolicious for CGI
20:39 yko maybe somewhere ill even finish it
20:39 sri who cares whats faster in a micro benchmark if your app is already fast enough ;p
20:40 yko it faster than Mojoliciouis about 25-30% in cgi mode, which i use often :(
20:41 yko i know you don't care about cgi at all :D
20:41 sri i do actually
20:41 xantus_ heh, I was suppose to upload a new version of Mojo::Server:FCGI
20:41 * sri cracks the whip
20:42 xantus_ my eye!
20:42 sri i'm more concerned about scalability and usability than raw performance these days
20:42 sri scalability != performance
20:43 yko i understand you, that is whu i don't cry about cgi perf. in #mojo last month :)
20:43 yko *why
20:43 yko well, one day ill finish it enough to publish it. there's some tricks. i cut off all websocket-related stuf, app->client, static, and packed all controllers in one file at least.
20:43 yko in will be quite dirty and you won't like it, i'm sure
20:45 sri doesn't sound like it's worth 25% ;p
20:45 yko haha
20:45 yko you don't need shebsockets in cgi mode, im shure
20:45 sri yea, but those don't cost much
20:46 yko ye
20:46 sri static costs a little
20:46 yko nested controllers and static cost little also
20:46 sri and template detection
20:46 yko in most cases client is also optional. those who really need that can do just ref($app)->attr(client => ....)
20:46 sri if i was micro optimizing for cg i would cut everything cached
20:47 sri *+i
20:47 yko and inherited controllers eating some time also
20:47 sri template caches also cost
20:47 yko hmmm. you're right
20:47 sri so much convenience lost ;p
20:47 sri just for 25% :S
20:47 yko when ill came to that poing ill pay attention on that
20:48 sri but mojolicious is still very much usable for plain cgi
20:48 yko actually i use it in cgi now
20:49 yko it's not very fast, but who cares about 0.6/request in cgi mode :D
20:49 sri all i want to worry about is how do i make it faster and easier to build cool web apps
20:50 yko sri++
20:50 yko that's why i use M9s instead of CGI even if it's 300% faster
20:51 yko whatever, i should go work, my children wanna eat
20:52 sri fact is there is always a tradeoff, convenience costs performance but gives you the agility to develop cooler apps
20:52 xantus_ heh, someone is sending spam and using my domain as the reply-to
20:53 xantus_ so I'm getting a bunch of bounces, etc
20:53 sri my spam filter eats those
20:53 xantus_ and one moron replied to the spam with their contact details
20:53 xantus_ lol
20:53 sri lol
20:54 xantus_ they spelled loan: lone
20:54 _mpu thus it works !!
20:54 sri well, they wouldn't keep spamming if it didn't work
20:54 * xantus_ nods
20:54 xantus_ I should be like, sure, I can get you that money, I just need a small fee to release the funds
20:54 xantus_ just western union me the money
20:54 _mpu their should have internet licenses
20:54 xantus_ lol
20:55 sri :D
20:55 xantus_ oh dear, she replied from an @aol account
20:56 xantus_ hahahah
20:58 yko oh! sri...
20:58 yko one thing i cut off in cgi mode is..... '
20:58 yko use base 'Mojo';
20:58 * yko giggles
20:59 * xantus_ ses yko on fire
20:59 xantus_ sets
20:59 yko damn, it's hot enough now! stop it!
21:00 xantus_ easier to set the fire
21:00 xantus_ like charcloth
21:01 yko :\
21:07 yko <3 Mojolicious
21:07 sri viva la revolucion!
21:07 yko xantus was right, that's just triangle with a butt...
21:08 yko xexe
21:08 sri
21:09 yko argh.. i should change default miranda font
21:13 yko much better. now i see when sri paints some butts with unicode characters :D
21:13 sri
21:13 xantus_ :P
21:14 xantus_
21:14 sri oh we could use that as a naming scheme for mojolicious, our favorite unicode characters
21:15 sri current one is snowman, which fits in
21:15 sri http://unicode.org/Public/6.0.​0/charts/versioned/U1F600.pdf
21:15 garfield http://xrl.us/bhtvix
21:16 sri "SMILING CAT FACE WITH HEART-SHAPED EYES" could be the next codename
21:16 yko damn.. DejaVuSans doesn't shows me your unicode :(
21:16 yko Myriad too
21:16 xantus_ ☄ is a comet
21:17 xantus_ deja vu works fine
21:17 yko sri what font do u use in im?
21:17 sri "Mojolicious 1.0 (ANGRY FACE) released"
21:17 sri hmm
21:17 xantus_ lol
21:19 vti
21:20 sri
21:20 vti call me
21:20 vti
21:20 sri Comet would be a good codename
21:21 vti phone is a good name too :D
21:22 yko ☀ looks like bug
21:22 vti
21:23 vti ☁☁☀☁☁
21:23 GitHub82 joined #mojo
21:23 GitHub82 mojo: master Sebastian Riedel * 0facbd7 (1 files in 1 dirs): added code name scheme to coding guidelines - http://bit.ly/cdXXRs
21:23 GitHub82 left #mojo
21:24 sri makes it easier to let the most active contributors decide the next code name too
21:25 yko haha
21:27 xantus_ cool
21:27 sri ANGRY FACE is not an option for 1.0 though ;p
21:28 xantus_ I would hope its not
21:29 vti ☁✈☄☠
21:29 sri not the skull...
21:29 vti it's a story!
21:30 vti sky plane comet... crash!
21:30 sri :o
21:30 sri ☄vti☠
21:32 vti sri ✘
21:33 vti
21:35 sri http://unicode.org/Public/6.0.0/charts/versioned/ # many good names in there :)
21:35 garfield [ Unicode 6.0 Versioned Draft Charts Index ]
21:37 vti Zzz...
21:41 sri good idea
21:41 sri Zzz
21:42 xantus_ I'M ON A BOAT
22:14 xantus_ hmm, maybe I should switch to mojox couchdb
22:33 xantus_ ugh, frustrating
22:53 xantus_ their docs could be better
22:58 und3f yees, spawn_fcgi is much better than Fcgi::Prefork

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