Camelia, the Perl 6 bug

IRC log for #mojo, 2011-09-24

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

All times shown according to UTC.

Time Nick Message
00:48 MojoGuest30 joined #mojo
00:48 MojoGuest30 From: http://t.co/xbW19y9q (22 hits)
00:50 MojoGuest30 testing
00:50 MojoGuest30 test
00:50 MojoGuest27 joined #mojo
00:50 MojoGuest27 From: http://t.co/xbW19y9q (23 hits)
00:50 MojoGuest27 test
01:27 AndroUser joined #mojo
01:29 AndroUser are there any examples on a restful ldap authentication with mojolicious?
01:30 AndroUser anybody???
01:30 purl anybody is, like, there
02:30 crab poor androuser, never finding anyone awake.
02:46 MojoGuest934 joined #mojo
02:46 MojoGuest934 From: http://www.google.com/url?sa=D&amp​;q=http://dev.xantus.org&usg=A​FQjCNF_rM6vpg9DT3eh4McOgWqn8mn5Xw (1 hits)
02:59 MojoGuest540 joined #mojo
02:59 MojoGuest540 From: http://t.co/xbW19y9q (24 hits)
03:57 MojoGuest882 joined #mojo
03:57 MojoGuest882 From: http://t.co/xbW19y9q (25 hits)
04:25 perlrocks Twitter: "sudo sh -c "curl -L cpanmin.us  perl - Mojolicious". Install Mojolicious on your CI Server and Webserver, ... http://t.co/VFzdaKep" --Baileypofnp http://twitter.com/Baileypof​np/status/117454596850462720
04:32 kaare joined #mojo
05:18 preflex_ joined #mojo
05:22 MojoGuest925 joined #mojo
05:22 MojoGuest925 From: http://t.co/xbW19y9q (27 hits)
05:44 * cakeftw yawns
05:44 * cakeftw still has no cake
05:55 Foxcool joined #mojo
06:15 Vandal joined #mojo
06:21 baton8 joined #mojo
06:32 cakeftw marcus: https://github.com/kraih/mojo/issues/221 # hmm, i would expect better written issues from him
07:11 MojoGuest584 joined #mojo
07:11 judofyr joined #mojo
07:26 crab | is in "unwise", right?
07:32 d4rkie joined #mojo
07:33 MojoGuest737 joined #mojo
07:33 MojoGuest737 From: http://paper.li/MiCHiLU/1302059249 (1 hits)
07:35 judofyr_ joined #mojo
07:45 marcus cakeftw: I would think this goes under being 'pragmatic', if browsers does not encode |
07:45 marcus cakeftw: same as we did for nginx spec transgressions
07:45 marcus safari dtrt tho so I'm not sure
07:45 marcus firefox does the same as chrome tho
07:47 cakeftw marcus: then the proposal is missing
07:48 marcus agree
07:48 judofyr_ oh, you're cake today
07:49 marcus he became cake since I got cake from hyper_linda
07:49 cakeftw my cover is blown :/
07:49 judofyr too bad I can't tab-complete on postfix
07:49 marcus coverftw
07:50 crab i wish you'd just go back to sri. i'm too old to adjust to sudden changes.
07:51 cakeftw heh
07:52 crab or...
07:53 sri now if i'm effusively super-polite and helpful, i can confuse a lot of people.
07:54 sri including myself. damn, it looks weird to see <sri> in front of things i say.
08:00 baton8 joined #mojo
08:02 marcus yay
08:02 marcus we can make a supersri!
08:03 marcus crab: chicken
08:03 noganex_ joined #mojo
08:03 crab wow, nginx 1.1.4 has support for persistent connections in the http proxy
08:24 cakeftw lol
08:39 * cakeftw goes to play with some toxic waste
08:39 supersri \o/
08:41 marcus we're having a hackathon this weekend
08:45 * supersri has just been playing with ways to improve the accept performance of hypnotoad
08:45 marcus MONGODB
08:46 marcus accept performance is nice too tho ;)
08:46 supersri managed a 1000% increase, but there are tradeoffs :S
08:50 crab sri: tell more?
08:50 supersri problem is the accept mutex
08:50 supersri all processes that don't have it nuke the whole listen socket watcher
08:51 supersri then once they get it they recreate it for one accept
08:54 supersri basically the switch between listening and not listening is way too expensive atm
08:55 crab ah.
08:57 supersri only affects EV though
08:57 supersri the IO::Poll backend doesn't care too much
09:24 GitHub72 joined #mojo
09:24 GitHub72 [mojo] kraih pushed 1 new commit to master: http://git.io/1MnbUg
09:24 GitHub72 [mojo/master] improved accept performance of the standalone daemon with EV backend by up to 1000% - Sebastian Riedel
09:24 GitHub72 left #mojo
09:25 supersri i'm sure there's a simple solution for hypnotoad too, but i just can't think of it atm :S
09:28 supersri prolly some EV trick to keep io watchers active without warching anything
09:33 supersri creating new io watchers is really fricking expensive
09:36 supersri https://gist.github.com/1239013 # wow, lua is *a lot* faster than v8 javascript
09:36 perlrocks Twitter: ""MCPC: Mojolicious on Google Chrome Web Browser: To view the Twitter stream in real time from the Lite server connected WebSocket" http://t.co/NwExcCTK" (ja) --CLCLCL http://twitter.com/CLCLCL/​status/117532791033442304
09:52 crab lua is a fairly pleasant language
09:56 judofyr joined #mojo
09:58 judofyr has anyone solved the store Mojo-JSON-booleans-in-MongoDB
10:14 mk joined #mojo
10:17 mk hi, is it safe in Mojolicious to use __FILE__ in the app?
10:17 mk for example $self->stash(file => __FILE__);
10:19 mk or is there any way in mojolicious to get app filename which is executed, in similar way like app->home ($self->app->home)
10:20 judofyr joined #mojo
10:20 mk in my test lite app I see $ENV{MOJO_EXE} is set to the file name, but it's not always there
10:31 judofyr_ joined #mojo
10:31 theartistformerlykn aww
10:34 tafkas tempire: it's tomorrow, you promised me a discussion about the future… or something like that
10:41 tafkas trigger, promises or futures?
10:44 perlrocks Twitter: "Testing with Mojo is great fun. The Mojolicious framework provides an easy way to test your web applications wit... http://t.co/D5cHiqsU" --Florapclsu http://twitter.com/Florapcls​u/status/117549937817944064
11:05 noganex joined #mojo
11:09 sri crab: where did you read about keep alive support in nginx?
11:19 sri found just a few mailing list posts that are not very informative
11:48 mk in mojolicious::lite is there any way to see what methods are under 'app' or '$self->app' ?
11:48 mk not looking into the sources, if I would like to start from documentation
11:49 mk I know there is app->start, app->home, but what else is there?
12:29 crab sri: i just saw it on nginx.org, second item from top
12:30 crab 1.1.4 release, "Support of persistent connections with upstream servers." i haven't looked further yet.
12:30 crab mk: man Mojolicious <- all those methods
12:37 sri ah
12:37 sri old implementations seem to only support memcached
12:38 sri but judging by the mailing list posts it looks like it works for all protocols
12:59 GitHub26 joined #mojo
12:59 GitHub26 [mojo] kraih pushed 1 new commit to master: http://git.io/VX7sQA
12:59 GitHub26 [mojo/master] improved accept performance a little more - Sebastian Riedel
12:59 GitHub26 left #mojo
13:03 stephan48 how does this inreac
13:03 stephan48 increase performance?
13:03 purl i guess increase performance is not always gauranteed
13:04 GitHub115 joined #mojo
13:04 GitHub115 [mojo] kraih pushed 1 new commit to master: http://git.io/yCqbdg
13:04 GitHub115 [mojo/master] fixed small Mojo::IOLoop bug - Sebastian Riedel
13:04 GitHub115 left #mojo
13:05 sri it actually broke something :S
13:05 * sri is fighting with side effects
13:06 sri oh well… enought micro optimizing for today
13:06 andrefs joined #mojo
13:25 andrefs joined #mojo
13:32 MojoGuest334 joined #mojo
13:32 MojoGuest334 From: http://t.co/xbW19y9q (2 hits)
13:43 MojoGuest334 hello
13:43 purl que tal, MojoGuest334.
13:43 MojoGuest334 how are you?
13:54 MojoGuest554 joined #mojo
13:54 MojoGuest554 From: http://t.co/xbW19y9q (3 hits)
13:59 janus hi
14:00 janus integrating mojo into xslt is nice... need to make than a proper module
14:01 janus makes it suprisingly comfortable, since its markup anyways
14:01 janus in general the current xslt module is a bit unwieldy imho, so i am looking to integrate it there
14:04 janus and i wanted to mention how much i appreciate Mojo::UserAgent :)
14:05 janus and ::DOM of course ;)
14:05 * janus has written quite a number of scrapers in the past few months
14:06 sri glad you like it :)
14:07 janus glad you wrote it :)
14:18 janus heh, any xsd schemas for mojo?
14:19 janus having a quite simple yet universal non-recursive xml serializer that renders the stash...
14:20 janus pretty cool since it allows to forget about the hairy stuff most of the time
14:21 janus taking namespaces into account would be very simple too - just mapping classes to urn:bla:my:package
14:23 janus so all needed is a bless, not really anything materialized - as in, could even work out of the box for ORMs
14:25 janus yet i still think about having a lazy sserializer, since i don't like the overhead - not sure if its even worth the complexity...
14:58 * crab ponders hairy stuff
14:58 abra joined #mojo
15:24 perlrocks Twitter: "Getting into Mojolicious http://t.co/uchqySMC" --klhughes http://twitter.com/klhughes​/status/117620434073157632
15:47 lammel2 joined #mojo
15:55 sri hmm
15:55 sri with multi-accept i can make the EV backend actually faster than the IO::Poll one at accepting
16:07 sri almost 1000% for hypnotoad too
16:28 gbacon joined #mojo
16:29 MojoGuest418 joined #mojo
16:34 mk crab: yeah, read that manual, but not all of the methods are there for Lite, I just wanted to be sure is there anything there to get filename of the app, something like __FILE__
16:45 GitHub25 joined #mojo
16:45 GitHub25 [mojo] kraih pushed 1 new commit to master: http://git.io/lxqkMQ
16:45 GitHub25 [mojo/master] improved accept performance of all built-in servers by up to 1000% with the EV backend - Sebastian Riedel
16:45 GitHub25 left #mojo
16:46 sri i need a windows tester for this change
16:56 sri interesting how hypnotoad with EV backend now beats starman without keep-alive
16:57 sri 700 vs 600 rps for the same lite app with "ab -n 10000 -c 100 http://127.0.0.1:5000/"
16:59 sri hmm
17:00 sri i forgot to disable ipv6 and tls support
17:00 sri then it jumps another 100 rps :)
17:00 sri 800 vs 600
17:01 sri and without EV it't just about 30 rps slower
17:01 sri 770 rps with pure perl
17:06 perlrocks Twitter: "Made hypnotoad quite a bit faster for non-keepalive connections today, testing would be appreciated. http://t.co/rnfRGjGY #perl #mojolicious" --kraih http://twitter.com/kraih/status/117646015569461248
17:19 sri 1200 vs 1100 rps for a super micro hello world without keep alive
17:19 marcus sri: neato
17:20 sri 1800 vs 1900 with keep alive
17:21 sri ops
17:21 marcus so slower than starman with keepalive?
17:21 marcus slightly
17:21 sri i forgot todisable optional xs again in hypnotoad
17:22 sri it's a little closer then
17:22 sri 50 rps or so
17:23 sri starman has more worker processes though
17:23 sri and xs parsers
17:24 sri even without any xs at all we don't drop below 1750 in this test :)
17:25 sri oh, just for the record, this benchmark is just a hello world without any relation to the real world
17:25 andrefs joined #mojo
17:25 sri i'm just looking for bottlenecks
17:26 sri (the kind nobody else would ever notice)
17:28 sri still not sure about multi-accept though
17:28 * sri is very much opposed to anything that adds more complexity
17:29 sri accepts with EV very scary slow though
17:30 sri as in below 100 rps slow
17:30 sri so it might be worth it
17:32 sri every time i optimize hypnotoad i get the urge to rewrite it from scratch -.-
17:33 crab why? i mean, what do you want to change?
17:33 stephan48 sri: so do it?
17:33 sri as in rip it out of mojolicious and make it a separate project ;p
17:34 sri and rewrite half of it in C :)
17:35 sri not because hypnotoad is bad, but because it would be much easier to optimize further
17:36 * sri loves hypnotoad
17:39 crab could create a separate C implementation
17:42 sri not like there is any chance i'll have the spare time to do it :)
17:51 sri fast enough is fast enough, i think actually higher level features like maybe diagnostics plugins should come first
17:54 mk hi guys
17:54 mk could you spare 5min for a noob?
17:59 stephan48 no but you may just ask your question... maybe someone answers
18:01 mk i did, it just I'm having inconsintend $ENV each time I run my simple, hello-world-like app
18:01 stephan48 please explain further
18:02 mk I'm not always geting MOJO_EXE from...
18:02 mk % foreach my $key (sort keys %ENV) { # and so on..
18:02 stephan48 why you need the filename?
18:02 mk for debugging purposes
18:02 mk as I have actually different problem with Plack
18:03 GitHub0 joined #mojo
18:03 GitHub0 [mojo] kraih pushed 1 new commit to master: http://git.io/7kaqXw
18:03 GitHub0 [mojo/master] simplify multi-accept logic - Sebastian Riedel
18:03 GitHub0 left #mojo
18:04 sri MOJO_EXE is strictly internal use only and not documented
18:04 mk I just want be sure that from my naive Plack::Builder router correct script is executed, as right now..
18:04 mk cool
18:04 mk and is it okay to use __FILE__ then?
18:04 mk in mojo apps
18:05 sri if it works
18:05 mk :)
18:05 mk well, it works
18:06 stephan48 somehow i dont get what you want to archive
18:06 mk I'm askign as I see $0 is not correct but __FILE__, is
18:06 mk stephan48: it long story, I will probably send email to the list
18:06 sri yea, $0 is evil
18:07 mk but just had this question about $0 and __FILE__
18:07 sri just don't use it unless you control the whole stack
18:07 mk sri: do you mean $0 ?
18:07 * sri nods
18:07 mk sri: k
18:08 mk thanks
18:20 MojoGuest185 joined #mojo
18:20 MojoGuest185 From: http://t.co/xbW19y9q (4 hits)
18:21 MojoGuest185 wow, really, websocket + irc? cool!
18:21 MojoGuest185 em...Hello! :)
18:24 MojoGuest341 joined #mojo
18:24 MojoGuest341 From: http://t.co/xbW19y9q (5 hits)
18:26 MojoGuest753 joined #mojo
18:26 MojoGuest753 From: http://t.co/xbW19y9q (6 hits)
18:36 Kulag joined #mojo
18:40 sri anyone feel like maintaining Mojo::IOWatcher::AnyEvent?
18:40 sri just hacked it together
18:45 perlrocks Twitter: "Hacked together an AnyEvent backend for #mojolicious, anyone free to turn it into a #cpan module. :) http://t.co/QE5ZFdNj #perl" --kraih http://twitter.com/kraih/status/117670900425428993
18:49 perlrocks Twitter: "sudo sh -c "curl -L cpanmin.us  perl - Mojolicious". Install Mojolicious on your CI Server and Webserver, too... http://t.co/PZCfpJd1" --Jeanettafzelt http://twitter.com/Jeanettafz​elt/status/117672057407414272
18:50 grim_fandango joined #mojo
18:51 perlrocks Twitter: "“@kraih: Hacked together an AnyEvent backend for #mojolicious, anyone free to turn it into a #cpan module. :) http://t.co/rLT3CALP #perl”" --sharifulin http://twitter.com/sharifuli​n/status/117672539978874880
18:55 marcus sri: I thought that was already in the core dist?
18:55 sri marcus: EV, not AnyEvent
18:55 marcus ah
18:56 sri personally i don't care too much about normal AnyEvent, but people keep asking
18:58 sri EV can also be a bit tricky to install, due to the compiler requirement
18:59 sri so it makes sense for those that are only interested in getting some AnyEvent module working
19:01 sri call me crazy, but i could swear it was easier to find maintainers for my modules back in the catalyst days
19:03 gbacon I'm trying to write what should be a simple self-guided study app (where each lesson has text followed by fill-in-the-blank, true-false, and multiple-choice questions) with Mojolicious but am having a hard time deciding how to organize its code
19:04 gbacon it seems like a given lesson series ought to have a single layout that knows how to plug in text and questions
19:17 MojoGuest28 joined #mojo
19:17 MojoGuest28 From: http://news.ycombinator.com/item?id=1277067 (1 hits)
19:21 MojoGuest133 joined #mojo
19:21 MojoGuest133 From: http://news.ycombinator.com/item?id=1277067 (2 hits)
19:21 MojoGuest133 snazzy!
19:22 MojoGuest133 /j #perl
19:23 marcus sri: crazy!
19:24 marcus sri: we didn't give individual people maintainer roles, we kept it all in the common svn and I did any needed releasing
19:24 marcus at least for some time
19:25 marcus sri: people mostly want to maintain things they actually use...
19:35 sri true
19:42 gbacon joined #mojo
19:44 perlrocks Twitter: "Sounds different use. RT @ bsiyo: Google: MCPC: Mojolicious on Google Chrome Web Browser:: Twitter stream in real time to display the connected server WebSocket from Lite: ..." (ja) --iR3 http://twitter.com/iR3/status/117685917547896832
20:01 perlrocks Twitter: "MCPC: Mojolicious on Google Chrome Web Browser:: http://t.co/7tSb2FiO to view Twitter streams in real time from the Lite server connected WebSocket" (ja) --yamada258 http://twitter.com/yamada25​8/status/117690029479968769
21:11 arthas joined #mojo
22:03 MojoGuest314 joined #mojo
22:03 MojoGuest314 From: http://www.google.com/url?sa=t&amp;source=w​eb&amp;cd=7&amp;sqi=2&amp;ved=0CFYQFjAG&amp​;url=http%3A%2F%2Fdev.xantus.org%2F&amp;rct​=j&amp;q=websockets%20demo&amp;ei=5VB-TtfEK​YLqgQeTm-VI&amp;usg=AFQjCNGw6nZv0wO42uzUnHh​WcD0vcKNULg&amp;sig2=fZ1Oct-Cf_YEQgIY4EllBA (1 hits)
22:03 MojoGuest314 hello
22:05 perlrocks Twitter: "MCPC: Mojolicious on Google Chrome Web Browser: To view the Twitter stream in real time from the Lite server connected WebSocket: Google C. .. http://t.co/ecK2wwbx # hb_chrome" (ja) --onoben_jp http://twitter.com/onoben_j​p/status/117721354781986816
22:09 perlrocks Twitter: "MCPC: Mojolicious on Google Chrome Web Browser:: http://t.co/Gj9NgUqh to view Twitter streams in real time from the Lite server connected WebSocket" (ja) --eiffig11 http://twitter.com/eiffig11​/status/117722273875636224
22:26 * tempire comes up for air
22:26 tempire snapshots of distributed systems are hard.
22:27 tempire I didn't realize until yesterday that mojo had a little event driven system in it
22:28 sri tempire: it's the magical pixie dust that makes everything in mojolicious work :)
22:28 tempire pipelined message processing is neat
22:29 sri as in http pipelining?
22:29 tempire er, I mean promise pipelining
22:30 sri oh
22:30 tempire what were you thinking about renaming?  ioloop::trigger?
22:32 sri ye
22:34 tempire ::trigger's current implementation doesn't look like promises to me.
22:35 tempire though you could implement promises on top of it, probably
22:36 sri it contains promises, but now you understand my problem ;)
22:43 mattastrophe joined #mojo
22:52 tempire so you're looking for an easier/more intuitive way to do what ::trigger does, I presume
22:53 sri ?
22:53 tempire I guess not.  I don't know what we're talking about, then.
22:54 * tempire stares at the clouds
22:54 sri i was looking for a more descriptive name
22:54 sri if you have an idea for making the api nicer, shoot, but i'm not looking
22:55 tempire ok
22:55 sri in fact i'm rather happy with the api
23:04 perlrocks Twitter: "MCPC: Mojolicious on Google Chrome Web Browser:: http://t.co/kxUOxPQ9 to view Twitter streams in real time from the Lite server connected WebSocket" (ja) --wakeari_maru http://twitter.com/wakeari_ma​ru/status/117736078953287680
23:26 andrefs joined #mojo

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