Camelia, the Perl 6 bug

IRC log for #mojo, 2012-08-02

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

All times shown according to UTC.

Time Nick Message
00:10 davido "Mojolicious/Guides/Routing.pod#Conditions" has a broken link to "Mojolicious/Routes.pm#over" that should direct to "Mojolicious/Routes/Route.pm#over"
00:11 sri thanks, will fix it
00:19 xaka1 sri: btw, there are many places where events, attributes and methods have the same name so #<name> always sends you to whatever comes first even if you are looking at different entity. probably each link should come with a prefix like #event-<name>
00:19 sri xaka1: you're welcome to fix it
00:21 xaka1 what do you use to generate HTMLs from PODs?
00:28 SmokeMachine_ joined #mojo
00:34 GitHub159 joined #mojo
00:34 GitHub159 [mojo] kraih pushed 1 new commit to master: https://github.com/kraih/mojo/commit/9f​795ac8e8ba04b2f0009a1bb3e5a154df77facb
00:34 GitHub159 [mojo/master] test default limits - Sebastian Riedel
00:34 GitHub159 left #mojo
00:57 laouji joined #mojo
01:00 zhutingting joined #mojo
01:02 scott joined #mojo
01:05 perlite joined #mojo
01:38 jnap joined #mojo
01:43 sri actually, there's an easy fix for the anchor problem
01:47 preflex_ joined #mojo
01:47 GitHub150 joined #mojo
01:47 GitHub150 [mojo] kraih pushed 1 new commit to master: https://github.com/kraih/mojo/commit/9c​89a67e5586337db6c81029f304fbe8426c77a7
01:47 GitHub150 [mojo/master] fixed bug that prevented multiple anchors with the same name in Mojolicious::Plugin::PODRenderer - Sebastian Riedel
01:47 GitHub150 left #mojo
01:49 sri and it's live :) http://mojolicio.us/perldoc/Mojo/UserAgent#start1
01:54 d4rkie joined #mojo
02:01 xaka1 joined #mojo
02:04 preflex joined #mojo
02:08 noganex_ joined #mojo
02:32 Liam`` joined #mojo
02:36 Psyche^ joined #mojo
02:37 GitHub35 joined #mojo
02:37 GitHub35 [mojo] kraih pushed 1 new commit to master: https://github.com/kraih/mojo/commit/53​ff820d467081f00f305a705876226be8e7cd10
02:37 GitHub35 [mojo/master] remove special characters from anchors - Sebastian Riedel
02:37 GitHub35 left #mojo
02:53 bird joined #mojo
03:12 phillipadsmith joined #mojo
03:13 chansen joined #mojo
03:34 jnap joined #mojo
04:11 spaceships joined #mojo
04:17 xaka joined #mojo
04:18 asarch joined #mojo
04:30 scott__ joined #mojo
04:40 dwierenga joined #mojo
04:46 asarch joined #mojo
04:51 lukep joined #mojo
05:16 ovnimancer joined #mojo
05:24 zhutingting joined #mojo
05:28 xaka1 joined #mojo
05:30 Britzel joined #mojo
05:47 Elraraniel joined #mojo
05:48 Elraraniel Dumb question: if I'm using perlbrew, will the mojo one-liner "curl get.mojolicio.us | sh" install to perlbrew correctly?
05:50 spleenjack joined #mojo
06:02 pau4o joined #mojo
06:02 Foxcool_ joined #mojo
06:29 Vandal joined #mojo
06:31 amirite dummy
06:34 Patterner joined #mojo
06:59 bc547 Elraraniel: yes
06:59 Elraraniel bc547:  Thanks :)
07:18 scott joined #mojo
07:30 taryk joined #mojo
07:39 fhelmber_ joined #mojo
07:42 amirite joined #mojo
07:50 Elraraniel left #mojo
08:03 cosmincx joined #mojo
08:47 mire joined #mojo
08:52 test joined #mojo
09:54 nicomen joined #mojo
10:07 loy joined #mojo
10:10 spleenjack1 joined #mojo
10:12 amirite joined #mojo
10:14 nic Aha!  My revelation for today... when using perlbrew, you don't need to suffer crazy-long paths
10:15 nic If you don't care about switching around which perl you use, you can do "perlbrew install -D prefix=/opt/perl perl-5.16.0"
10:16 nic It then provides you /opt/perl/bin/perl etc
10:20 crab if you don't care about switching around which perl you use, what good is perlbrew really?
10:20 crab i guess it does make it a lot more convenient to install $version of perl.
10:21 bc547 crab: some distro's do not have a recent enought perlversion.. that's where perlbrew comes in (without screwing up the whole system)
10:22 crab bc547: what i meant is that you can always install that specific new version of perl to ~/perl by hand if you want... but yes, perlbrew makes it easy as pie.
10:32 daxim joined #mojo
10:46 nic bc547, crab: The route I suggested is v good if you want to switch perls.  Turns out it puts stuff under /opt/perl/lib/5.16.0
10:46 nic trouble is, that's all it's good for
10:46 Lucas1 joined #mojo
10:46 nic for my non-dev boxes I'm returning to the traditional route
10:47 nic perlbrew gets a bit tangled about env vars, so the traditional route is actually easier
11:09 crab it's always a bit dicey whether you have things set up properly that cron jobs will find the rightly-brewed perl and such.
11:10 nic yeah, with perlbrew I really dislike the reliance on env vars
11:11 ovnimancer joined #mojo
11:20 * sri likes the way perlbrew works
11:20 sri if you have only one perl version on a dev box you're doing it wrong
11:22 sri if you don't want to support legacy perls that's fine, but there are always two versions of perl that are still maintained by the community and you should test on
11:22 * nic has three perl versions
11:23 sri also being able to test with and without threads can be very useful
11:23 nic that was my original motivation
11:50 crab i actually use perlbrew in production too. just for convenience of setting up something other than the system perl.
11:58 taryk joined #mojo
12:07 marcus sri: I'm wondering, if you do async in the mojo startup, is there any way to guarantee that they are done before the app starts accepting connections?
12:09 sri marcus: no
12:09 sri async is async
12:15 sri marcus: as a matter of fact, the event loop is not actually running during startup
12:16 marcus sri: interesting. so I could actually use Mojo::Delay during startup
12:16 marcus Mojo::IOLoop::Delay even
12:16 sri and $delay->wait?
12:16 marcus yes
12:16 sri that can mess up EV
12:17 marcus \m/
12:17 marcus how come?
12:17 marcus I was actually going to switch to use redis EXEC/MULTI instead anyways.
12:17 sri EV is a little unpredictable when it comes to initialization
12:18 sri if you initialize it too early, say in a hypnotoad manager process, it's hard to say if it will survive fork
12:18 marcus is there a hook to do something once the ioloop is running?
12:19 sri as you know, epoll/kqueue can't be inherited by forked children
12:19 marcus I guess I could just queue everything and then let it start, and run the callbacks
12:19 sri marcus: a 0 timer of course :)
12:19 marcus mm
12:19 sri yes, i strongly recommend 0 timers
12:20 sri fork safety with event loops is a nightmare
12:21 marcus mm
12:21 sri especially if we want to support more backends in the future i believe the only good solution is to not start the event loop at all in the manager process
12:22 sri that's tricky with I/O watchers though
12:22 sri you're welcome to investigate the problem and come up with a plan, but it is a very hard problem
12:26 sri marcus: currently, if you use a new Mojo::IOloop instance, or make a blocking Mojo::UserAgent request during startup you should be fine though
12:27 sri because we initialize the singleton event loop very very early, and the IO::Poll one is very fork safe
12:27 ryozi joined #mojo
12:28 sri basically, make sure you have a poll reactor, then you're fine
12:28 taryk joined #mojo
12:30 sri once we add the new Mojo::Reactor backend with support for many more event loops the fork problem could cause some very interesting bug reports :S
12:34 marcus yay
12:51 d4rkie joined #mojo
12:54 sri http://mojolicio.us/perldoc​/Mojolicious/Controller#req # when you see that description, what would you expect the return value of $c->req(Mojo::Message::Request->new) to be?
12:54 * sri is not sure how to handle aliases
12:54 sri or would you expect it to handle arguments at all?
12:54 crab i do not expect it to handle arguments
12:55 sri i see
12:55 crab and changing the request object is bizarre enough that i see no need to make a convenient shorthand to do it.
12:55 crab within a controller
12:56 sri crab: lets say $c->res(Mojo::Message::Response) then :)
12:56 sri umm
12:56 sri ->new
12:56 sri ok, my example sucks
12:56 sri for controller aliases it's pretty simple...
12:57 crab yeah, where's the problem?
12:57 sri http://mojolicio.us/perldoc/Mojo/Message#write
12:57 sri that's an alias that requires arguments
12:58 sri return values is undefined, but it will return the Mojo::Content instance
12:59 sri screw this... i'll document all aliases properly as methods -.-
13:04 rem_lex joined #mojo
13:19 jnap joined #mojo
13:51 CromeDome joined #mojo
13:56 gryphon joined #mojo
14:01 inokenty joined #mojo
14:16 GitHub196 joined #mojo
14:17 GitHub196 [mojo] kraih pushed 1 new commit to master: https://github.com/kraih/mojo/commit/ae​7829df5b3c89b50e391d4feae9278cb5befa53
14:17 GitHub196 [mojo/master] fixed dynamic content generation bug in Mojo::Message - Sebastian Riedel
14:17 GitHub196 left #mojo
14:23 sandeepdb joined #mojo
14:24 memowe \o/
14:24 sri \o\
14:26 memowe What's up here? I was busy with my Java (:/) data mining astronomy project. Planning to build a web frontend with Mojolicious and Inline::Java. :)
14:27 crab data mining astronomy project(!)
14:27 crab do tell.
14:29 memowe We classify objects from the Sloan Digital Sky Survey DR6 as quasars or post starburst galaxies.
14:29 memowe By "looking" at their spectra.
14:30 memowe That way it was possible to find some "false" preclassifications from that catalog. Also we helped astronomers to find new interesting galaxies - feels like real science. :D
14:30 sri /o/
14:30 phillipadsmith joined #mojo
14:31 memowe A cool follow-up project would be a ui where astronomers can "draw" spectra and build a classifier from that drawing.
14:32 memowe We use unsupervised machine learning here. 1NN and SVM with Gaussian kernel work great.
14:46 marcus bridges don't work well with async :(
14:47 marcus guess I need to use around_dispatch;
14:47 marcus or maybe not.
14:47 sri async bridges would suck
14:47 marcus sri: how do you suggest authenticating against the redis driver then?
14:48 marcus actually, I guess I don't need to lookup in redis, just in the session.
14:48 sri no idea, doing that stuff async sucks
14:49 sri why would you go async in the first place? redis is like super fast
14:50 sri i wouldn't be surprised if it actually costs you performance
14:52 sri not only is redis super fast, it is also single threaded, that means parallel queries don't gain you anything besides more memory usage
14:53 * sri shrugs
14:55 marcus sri: you need to do async to do stuff like redis pubsub in websockets tho?
14:56 pau4o left #mojo
14:56 sri pubsub is a valid use
14:57 marcus using two different redis drivers seemed a bit silly to me.. Although it might be a lot easier.
14:58 marcus melo's redis module seems quite solid.
14:58 sri convert the driver to a blocking/non-blocking hybrid?
14:58 sri that requires a lot of insight into how event loops work internally though
14:59 marcus yeah, that some scary shit :)
15:00 marcus think I'll focus on writing actual app code for now...
15:00 marcus too much yak shaving lately
15:08 GitHub151 joined #mojo
15:08 GitHub151 [mojo] kraih pushed 2 new commits to master: https://github.com/kraih/mojo/com​pare/ae7829df5b3c...9d1ef44f33b9
15:08 GitHub151 [mojo/master] better descriptions for a few methods - Sebastian Riedel
15:08 GitHub151 [mojo/master] better synopsis examples for message classes - Sebastian Riedel
15:08 GitHub151 left #mojo
15:23 vel joined #mojo
15:35 GitHub188 joined #mojo
15:35 GitHub188 [mojo] kraih pushed 1 new commit to master: https://github.com/kraih/mojo/commit/41​86f5e7a059aaacdd9533170cce61a6cfcc08ff
15:35 GitHub188 [mojo/master] better hook description - Sebastian Riedel
15:35 GitHub188 left #mojo
15:40 Leandr joined #mojo
16:09 xaka joined #mojo
16:11 alester joined #mojo
16:13 rwstauner joined #mojo
16:18 labrown joined #mojo
16:20 Lucas1 joined #mojo
16:21 Leandr joined #mojo
16:38 mire joined #mojo
18:24 Britzel joined #mojo
19:04 mire joined #mojo
19:50 gryphon joined #mojo
20:16 alxt joined #mojo
20:20 asarch joined #mojo
20:31 xaka joined #mojo
20:45 jwang joined #mojo
20:52 mire joined #mojo
21:24 mjy joined #mojo
21:47 lukep joined #mojo
22:15 jnap_ joined #mojo
22:46 alester_ joined #mojo

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