Camelia, the Perl 6 bug

IRC log for #mojo, 2010-08-10

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

All times shown according to UTC.

Time Nick Message
00:00 und3f right UTF-8
00:00 und3f i just tried to do lc on that :)
00:01 und3f but why?
00:01 und3f default charset for Mojo::Message is UTF-8
00:02 sri ?
00:02 und3f i got message Unknown encoding 'UTF-8;' at /home/und3f/projects/mojo/lib/Mojo/ByteStream.pm line 449
00:03 und3f oh, i see
00:03 und3f there is ending ;
00:10 perlrocks Twitter: "RT @kraih RT @tempiretech: Accessing Google Voice with #perl and #mojolicious. http://bit.ly/aZ6pyF" --zakame http://twitter.com/zakame/status/20748717757
00:11 xantus_ LOL
00:11 xantus_ it actually makes you pie
00:13 xantus_ π
00:17 tempire It's true.
00:21 GitHub102 joined #mojo
00:21 GitHub102 mojo: master Sebastian Riedel * 6a06bfc (1 files in 1 dirs): fixed typo - http://bit.ly/aNVDPe
00:21 GitHub102 left #mojo
00:34 und3f sri, found bug in dom building from response
00:34 und3f if content-type is like "application/atom+xml; charset=UTF-8; type=feed" send you pull request with fixups
00:35 sri und3f: you know the drill, make test case and it will get fixed ;)
00:35 und3f sri, i did
00:35 und3f and fixed
00:35 sri und3f++
00:43 GitHub41 joined #mojo
00:43 GitHub41 mojo: master und3f * 0e57655 (2 files in 2 dirs): Fixed encoding picking for dom building
00:43 GitHub41 mojo: master Sebastian Riedel * 62aa1c1 (2 files in 2 dirs): cleanup
00:43 GitHub41 mojo: master commits 6a06bfc...62aa1c1 - http://bit.ly/dnVczE
00:43 GitHub41 left #mojo
00:43 und3f \o/
00:45 ashleydev joined #mojo
00:47 sri i kinda like the idea of $loop->write($id => $chunk);
00:47 sri but the low watermark callback required ruins it
00:49 ask joined #mojo
01:37 stephan48 joined #mojo
01:39 go|dfish joined #mojo
01:39 doubi joined #mojo
02:05 chilts joined #mojo
02:41 Foxcool joined #mojo
02:59 Alias joined #mojo
03:32 janus joined #mojo
03:43 yakudzo joined #mojo
04:04 tholen42 joined #mojo
04:45 dynax60 joined #mojo
05:12 koban joined #mojo
05:28 koban left #mojo
05:28 koban joined #mojo
05:55 yko joined #mojo
06:02 sri damn, it's crazy how quick my req/s drop from 20k back to 12k
06:06 sri oh well, i better move on, this is really fun but there's no actual value in it
06:19 ysyrota joined #mojo
06:26 yko ап
06:29 Kai_ joined #mojo
06:29 Kai_ Why does my websocket get disconnected after 5 minutes?
06:30 sri timeout
06:30 Kai_ what do I do?
06:31 Kai_ websocket_timeout, I see.
06:31 sri ye
06:31 sri there is no command line option for it yet hmm
06:33 sri give me a minute :)
06:34 Kai_ let's see, i'm making the client ping every 200s
06:34 Kai_ will that work?
06:34 Kai_ and by ping I mean send something through the websocket.
06:36 GitHub78 joined #mojo
06:36 GitHub78 mojo: master Sebastian Riedel * 3b687a4 (2 files in 1 dirs): added command line option for websocket timeout - http://bit.ly/c6YYvO
06:36 GitHub78 left #mojo
06:36 sri Kai_: there ya go :)
06:37 Kai_ :)
06:46 sri hmm, in tornado they use $loop->write($chunk, sub {...})
06:48 Htbaa joined #mojo
06:49 sri dunno, passing around even more callbacks doesn't look like the answer
06:53 sri guess this is one of the cases where we'll have to gather more experience first before a decision can be made
06:54 sri can always redesign the ioloop for 2.0 :)
06:55 jeff joined #mojo
06:57 jeff I am getting Can't locate object method "load_build" via package "Mojo::Loader" at /usr/local/lib/perl5/site_​perl/5.8.9/Apache2/Mojo.pm line 36
06:57 Kai_ sri: tornado?
06:57 jeff anyone know if there's a better version of Apache2::Mojo
06:58 jeff I have latest CPAN
06:58 sri Apache2::Mojo is dead code i think
06:58 jeff yikes... what should I be using?
06:58 sri i would just switch to PSGI for mod_perl
06:58 jeff is there a url showing how to set that up
06:58 sri (if it has to be mod_perl)
06:59 sri i have a few cookbook recipes for fastcgi and stuff
06:59 sri all on github
06:59 sri http://github.com/kraih/mojo/blob/mast​er/lib/Mojolicious/Guides/Cookbook.pod
07:01 jeff just for development... can it be executed as a CGI script?
07:01 sri ?
07:02 sri PATH_INFO=/foo ./myapp.pl
07:03 vti there was an embedded dev server somewhere... :)
07:03 sri every mojolicious application *is* a cgi/fastcgi/server out of the box
07:03 sri ./myapp.pl daemon
07:03 jeff ok... so I can just set DocumentRoot and ScriptAlias and it'll work, right
07:04 sri cgi detection in the cpan version works mostly, the one on github should work everywhere though
07:08 GitHub160 joined #mojo
07:08 GitHub160 mojo: master Sebastian Riedel * ace22a6 (2 files in 2 dirs): new codename! - http://bit.ly/dxiEq8
07:08 GitHub160 left #mojo
07:09 sri :)
07:10 sri comet melts the snowman
07:11 perlrocks Twitter: "Comet melts the Snowman. http://bit.ly/9JfTjg #mojolicious #perl" --kraih http://twitter.com/kraih/status/20775651127
07:16 chansen joined #mojo
07:18 jeff when I set up Apache/CGI as on the cookbook pod... I get a message
07:19 jeff Tip: CGI, FastCGI and PSCGI environments can be automatically detected very often and work without commands.
07:19 jeff does that mean, it's not detecting and I should grab the github version
07:20 jeff or can I set some environment var to let it know I'm in CGI environment and not mod_perl
07:22 sri yes that means detection is not working
07:22 sri github should just work
07:22 jeff ok... what do I need to clone from github
07:22 sri or you could just modify your app->start call to app->start('cgi')
07:22 jeff oh... cool
07:23 jeff for now that'll work
07:23 sri it's like line 1 in the tutorial *cough* ;p
07:23 jeff sorry man... I'll take a look at that :)
07:23 sri perldoc Mojolicious::Lite
07:28 sri btw. getting a github version is usually as easy as git clone git://github.com/kraih/mojo.git
07:38 tholen joined #mojo
07:40 perlrocks Twitter: "There was not much love for Mojolicious." (ja) --nijisaigai5518 http://twitter.com/nijisai​gai5518/status/20776939283
07:40 sri aww
07:43 sri !twitter @nijisaigai5518 ♥
07:43 sri ops
07:43 sri !tweet @nijisaigai5518 ♥
07:43 perlrocks sri: http://twitter.com/perlrocks/status/20777084526
07:44 sri where is the love?
07:46 janus flexo stole it
07:47 sri !tweet flexo has stolen the love!
07:47 perlrocks sri: http://twitter.com/perlrocks/status/20777254101
07:47 janus :)
07:50 dotan joined #mojo
07:55 dynax60 joined #mojo
08:18 sri http://github.com/yko/mojox-cpan-uploader/com​mit/bc146a34324d2f7249f29a3c4a999be43c42321b
08:18 sri lol at the commit message
08:18 sri yko++
08:18 janus voodoo++
08:19 janus i didn't know of the commit needle though :)
08:19 sri http://github.com/yko/mojox-cpan​-uploader/blob/master/Changes#L7 # hahaha
08:22 * vti cracks a whip
08:33 yko hehehe
08:33 yko he really did that :'(
08:36 perlrocks Twitter: "RT: @kraih: Comet melts the Snowman. http://bit.ly/9JfTjg #mojolicious #perl" --sharifulin http://twitter.com/sharifulin/status/20779385077
08:37 Htbaa joined #mojo
08:40 perlrocks Twitter: "Comet melts the Snowman. http://bit.ly/9JfTjg #mojolicious #perl (via @kraih)" --vtivti http://twitter.com/vtivti/status/20779549959
08:41 marcus oh oh
08:41 marcus sri: so when is comet going to CPAN?
08:42 sri soon™
08:43 sri or in Blizzard terms... "when it's done"
08:43 yko comet goint to be 1.0?
08:43 marcus adding the client to $client project now
08:44 sri comet is 1.0beta
08:44 yko hm
08:44 yko :))
08:45 yko i expected something like that :D
08:45 vti when did i miss alpha? ;p
08:45 yko las 2 years? :D
08:46 yko *last
08:49 marcus seems http://empireenterprises.com/blogs/g​oogle_voice_with_perl___mojolicious has some unicode issues
08:49 sri yea
08:49 sri looks like double encoding
09:41 cognominal joined #mojo
09:43 mvuets joined #mojo
10:08 koban joined #mojo
10:10 Htbaa joined #mojo
10:12 sri any thoughts about $loop->write_cb(sub {...}) vs $loop->write($chunk, sub {...})?
10:15 vti does it mean that a callback is called when chunk is sent?
10:16 sri after the chunk is entirely sent
10:16 sri it's a low watermark callback (kinda)
10:17 sri somehow an async mainloop needs to be able to pull more chunks from somewhere
10:17 sri $loop->write($chunk, sub { shift->write($chunk2, sub {...}) })
10:18 sri thats basically it
10:19 sri thats the more direct alternative to our $loop->write_cb(sub {...})->writing($id)
10:19 sri actually it's $loop->write($id, $chunk, sub {...})
10:20 sri you trade the ->writing/not_writing for a finite callback chain
10:25 sri performance wise there is no big difference
10:25 sri it's a usability question
10:36 marcus hmm
10:44 perlrocks Twitter: "Cheryls glam and is full of mojolicious stuff xx" --Queen_Cheryl http://twitter.com/Queen_Cheryl/status/20785063479
10:45 marcus hahaha
10:45 sri Oo
10:46 yko bgg
10:47 yko *report for spam
10:47 sri *report for report
10:48 yko *report for bot!!
10:50 sri in soviet russia bot reports you!
10:52 yko in soviet russia you report yourself :p
10:53 marcus in california you can always find a party. In soviet russia, party can always find you.
10:53 yko bwahaha :)
10:53 yko rly
10:53 marcus that's an original soviet russia joke.
10:54 marcus by Yakov Smirnoff.
10:56 vti there are no computers in soviet russia
10:58 yko ye, we justhave sockets beneath ears
10:59 mvuets They are. Because of no sex, you actually have to f*ck something.
11:00 yko Rubik's Cube, Tetris. What more do you need?
11:00 yko Go dig the garden :p
11:01 sri is anyone actually using raw Mojo::IOLoop?
11:01 yko not me. I'm interested in, but very lazy
11:02 yko You plan to change the api?
11:02 sri the ->write thing would break it :/
11:02 mvuets Lazy as in "lazy evaluations"? (:
11:02 yko Lazy as i am. Thants more than enough
11:02 sri it basically changes the whole flow
11:03 sri i know xantus is using ioloop, but he asked for ->write, so i doubt he would mind a little breakage ;p
11:04 yko btw, mvuets, did you read all 55 comrad Lenins' books? (and hidden bonus - personal workbook)? :)
11:04 yko sri: i don't think there's much peolpe who use raw IOloop at the moment
11:04 sri sadly there is nothing that could be deprecated, it changes the core functionality
11:05 yko but if you will wait few weeks - there will be :)
11:05 sri vti and xantuns only most likely
11:05 yko well... vti will suffer...
11:05 sri yko: i'm afraid of that
11:05 sri if i wait till after 1.0 people will get upset
11:05 yko ye, be shure
11:05 sri vti: thoughts?
11:05 yko i will be first
11:06 sri xantus: thoughts?
11:06 yko mb you can post message in google group and twitter
11:06 yko and wait for 1 day :)
11:06 DaTa break it!
11:06 * yko giggles.
11:07 sri yea, as soon as i start discussing this publicly someone will make a blog post about mojolicious braking backwards compatibility xD
11:07 sri *+e
11:08 * marcus fires up marsedit
11:09 sri guess a list post would be appropriate
11:09 marcus 'OMG! HE'S DOING IT AGAIN!'
11:09 yko hehe
11:09 sri :,(
11:09 yko oh, i forget: und3f uses ioloop i think
11:09 marcus good post topic?
11:10 yko damn, just break it! it' bettter to break until 1.0 then after
11:10 marcus just be sure it's a final change?
11:10 yko :))
11:10 yko marcus: you really think it's possible?
11:11 marcus sri: to be clear, this lets you send chunked output?
11:11 marcus yko: I mean, that he won't want to change it back after a release
11:11 sri marcus: it won't allow anything we can't do already
11:11 marcus not that he'll never change anything again...
11:11 marcus sri: so what's the point?
11:11 sri usability
11:11 marcus prettier api?
11:11 sri ye
11:12 sri doing stuff the right way
11:13 sri it might allow some extra optimizations, but that still needs to be investigated further
11:13 sri like avoiding a few syscalls for chunks small enough to fit into the kernel send buffer
11:14 sri i've been playing with that in my 20k req/s experiment
11:16 sri in the current design i can't explore those areas
11:18 sri baah...i don't know what to do
11:21 sri potential gain from those optimizations could be huge
11:22 yko break! break! break! :)
11:23 Foxcool joined #mojo
11:29 marcus sri: did you do a list post?
11:29 sri not yet
11:29 marcus http://www.huffingtonpost.com/adam-green/​breaking-google-goes-evil_b_676021.html?u​tm_source=feedburner&utm_medium=twitt​er&utm_campaign=Feed:+Tputh+(TPUTH+-+​Breaking+News+With+The+Social+Hammer) # Google went bad!
11:38 vti sri: i use only tick_cb
11:38 sri vti: thats unaffected
11:39 sri write_cb/writing/not_writing
11:39 vti no low-level stuff as for me
11:39 vti well, i had jabber client in mojo::ioloop..
11:39 sri writing/not_writing will be automatically decided internally
11:40 vti nobody will notice, i am sure
11:40 sri hmm
11:40 sri fingers crossed
11:56 marcus mst will notice ;)
11:57 vti :D
11:59 mvuets yko: no, i didn't (:
12:00 sri marcus: the big question is what would be the right way to handle such a change
12:02 vti sri: don't forget to put in Changelog "IOLoop cleanup" ;p
12:03 marcus sri: do a blog post?
12:04 marcus make it into a philosophical debate ;-)
12:04 sri heh
12:05 sri i think the change only affects the person that actually proposed it xD
12:06 Htbaa joined #mojo
12:07 sri honestly, i don't think debating the change would do any good
12:07 sri i'll explain it on the list and that will be it
12:08 sri and i hope xantus won't be too mad that i accepted his proposal :)
12:09 marcus it's pretty low level
12:10 sri about as low as it gets
12:11 sri the api is entirely unavailable in normal mojolicious apps
12:11 vti *normal*
12:11 vti hahahaha
12:16 marcus $self->client->ioloop->write ? :)
12:16 doubi joined #mojo
12:26 sri shoo!
12:27 sri ok, who wants a sleep ioloop api?
12:27 sri o/
12:27 sri hands in the air
12:28 sri *sleek
12:28 sri grr
12:28 marcus I want a sleeper cell
12:28 * sri plants a sleeper cell
12:29 marcus sri: seen rubicon?
12:30 vti sleep?
12:30 vti what do you mean by that?
12:30 marcus he means sleek
12:30 vti oh
12:32 sri ok, after porting the first few tests i'm officially in love with the new api
12:33 sri and willing to fight for it!
12:36 lbr hammertime
12:41 marcus apiphiliac
12:43 marcus now, tell us where the bad api designer touched you...
12:55 * lbr points
12:55 * sri paws marcus
13:13 sri hmm, this might double websocket performance
13:13 sri since writes are able to bypass the mainloop
13:15 sri small messages can go directly into the kernel send buffer
13:15 sri lets see
14:28 mvuets joined #mojo
14:44 ask joined #mojo
14:57 sri looks like this will be about 10% better performance, just by changing the api :)
14:58 sri and again open the door for much more radical optimizations
14:59 sri considering how low level this change is i'm quite surprised
15:26 dotan sri, do you do conferences?
15:27 sri nope
15:27 koban joined #mojo
15:32 dotan So who's the go-to Mojolicious speaker I should demand for future perl conferences? Sharifulin? vti?
15:32 sri sharifulin is awesome i heard
15:33 sri dunno if vti does conferences
15:33 dotan I know he was part of the Perl::Staff team with Gabor (at Cebit? Fosedam? some non-perl conference)
15:34 rindolf joined #mojo
15:35 dotan Right. I'm filling in my YAPC::EU survey, asking for Anatoly Sharifulin and Damien Conway as future speakers :)
15:35 rindolf left #mojo
15:37 vti i do conferences, but i really suck
15:38 dotan vti: will you come to Riga next year?
15:39 yko Hmm. Riga? Sounds good
15:39 yko vti, lets go Riga? :)
15:40 vti dotan: i can't plan that far =/
15:41 dotan I decided to go to Pisa the Thursday before it started :)
15:41 vti i wanted to go too, but something stopped me =/
15:44 vti dotan: sharifulin's english is not very good
15:50 dotan vti: I suspected that might be the case.
15:51 dotan I was thinking of writing something about mojolicious for Perl Vogue
15:51 vti dotan: i think that everybody waits for sri... with tomatos or eggs :D
15:51 dotan http://perlvogue.com/ http://www.slideshare.net/davorg/perl-vogue
15:52 dotan vti: If he'd give a talk we could probably expect a 50-minute API freeze ... :)
15:52 vti the biggest conference fight in history :D
15:53 vti dotan: is it a magazine/
15:53 vti ?
15:54 dotan Dave Cross gave a talk (2nd link), then came up the next day to say it wasn't a joke, and he'd really like to get articles for the site...
16:15 vti dotan: good idea
16:17 tempire I fixed my unicode issue months ago.  Unfortunately, I can't remember how I fixed it.
16:18 vti months is too long for mojo
16:28 tempire So my encoding is fine when dumped to the screen, but not when passed through render_json
16:35 ashleydev joined #mojo
17:05 ask joined #mojo
17:06 marcus dotan: I am coming to riga, and plan to submit a mojolicious talk.
17:08 marcus I submitted one for nordic perl workshop as well, but the conference was canceled because of volcanoes
17:08 sekimura_ joined #mojo
17:13 MojoGuest409 joined #mojo
17:16 MojoGuest358 joined #mojo
17:17 marcus but I'd love to see sri come to a conference some time.
17:17 * marcus is going to germany on friday. Frankfurt
17:18 tempire marcus: I feel like you should relish saying "the conference was canceled because of volcanoes".  Life just doesn't give that many opportunities to say such a thing.
17:18 marcus hopefully
17:19 MojoGuest282 joined #mojo
17:20 MojoGuest282 test
17:49 dotan marcus: cool!
17:49 dotan (also the volcanos bit is cool, but nm)
18:02 dotan left #mojo
18:11 xantus_ apparently, $work is claiming ownership of ignite
18:12 crab oh no. now what?
18:15 vti living on the edge!
18:18 spleenjack joined #mojo
18:24 yko ну вот, по крайней мере тест на компиляцию будет
18:25 * yko AFK
19:11 MojoGuest56 joined #mojo
19:11 MojoGuest56 Hi
19:12 garu hi MojoGuest56
20:19 MojoGuest816 joined #mojo
20:28 dotan joined #mojo
20:31 dotan left #mojo
20:31 dotan joined #mojo
20:37 koban joined #mojo
20:51 marcus xantus_: are you in trouble with $work for having open sourced ignite?
20:52 xantus_ no
20:52 xantus_ I started it on my own time
20:52 xantus_ and they wanted a demo asap, so I worked on it @ work
20:53 xantus_ ...and suspiciously, the vpo handed me invention assignment paperwork yesterday
20:54 xantus_ I made the repo private for now
20:55 marcus probably smart.
21:03 xantus_ I'll redesign it, and rewrite it
21:05 xantus_ check out CA labor code section 2870-2872
21:06 xantus_ http://www.leginfo.ca.gov/cgi-bin/displaycode?sec​tion=lab&group=02001-03000&file=2870-2872
21:43 dotan joined #mojo
22:06 doubi joined #mojo
22:08 ltriant joined #mojo
22:40 doubi joined #mojo

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