Camelia, the Perl 6 bug

IRC log for #mojo, 2011-10-03

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

All times shown according to UTC.

Time Nick Message
00:15 gshank joined #mojo
00:44 metaperl in Mojolicious::Lite, how do you get all the request parameters into a hash or hashref?
00:45 metaperl must be ->to_hash
00:51 metaperl no that didnt work
00:51 irctc039 joined #mojo
00:53 zenbitz joined #mojo
01:03 metaperl OK. I did it the long way - http://pastebin.com/9fPigcsB  .. is there a simpler way?
01:16 crab req->params->to_hash
01:28 sri this is ridiculous
01:29 sri like playing whack-a-mole
01:29 sri now it works on 5.14 and 5.12 but not 5.10 -.-
01:30 sri somehow the object destruction/ref counting/weaken code must have changed between those 3 versions
01:31 zenbitz joined #mojo
01:43 abra joined #mojo
02:01 crab moles!
02:11 D4RK-PH0ENiX joined #mojo
02:51 MojoGuest158 joined #mojo
02:51 MojoGuest158 From: http://news.ycombinator.com/item?id=1277067 (2 hits)
03:13 crab it's really too bad Mojo::Base isn't taint-proof
03:52 zenbitz joined #mojo
04:15 zenbitz joined #mojo
05:04 Eugene joined #mojo
05:35 koban joined #mojo
06:01 sromanov joined #mojo
06:13 Eugene joined #mojo
06:19 Eugene joined #mojo
06:21 Eugene joined #mojo
06:30 cosmincx joined #mojo
06:42 tabbi joined #mojo
06:42 tabbi left #mojo
06:51 tabbi joined #mojo
06:56 ysyrota joined #mojo
07:14 Vandal joined #mojo
07:15 abra joined #mojo
07:34 marting joined #mojo
07:48 koban joined #mojo
08:27 preflex_ joined #mojo
08:32 Eugene joined #mojo
09:00 spleenjack joined #mojo
09:15 tabbi left #mojo
09:49 sugar joined #mojo
11:05 AmeliePoulain joined #mojo
11:35 shonorio joined #mojo
11:39 trone_ joined #mojo
11:41 perlrocks Twitter: "@kraih removing #perl 5.8 from mojo kills it for rhel5 and centos5. #redhat do suck when it comes to perl, but worth considering" --PerlDean http://twitter.com/PerlDean​/status/120825819638669312
12:02 koban left #mojo
12:23 shonorio joined #mojo
12:29 sri damn you ref counting!
12:29 sri i can't find this stupid bug
12:48 marcus yeah
12:48 marcus perl needs moar bettar gc
12:55 judofyr joined #mojo
12:56 sri MOAR GC
12:56 trone joined #mojo
12:57 judofyr GC?
12:57 purl rumour has it GC is the boehm conservative garbage collector at http://reality.sgi.com/boehm/cg.html or a really really bad perl "programmer" or GrandCentral.com or branches/gsoc_pdd09 or a travesty against god
12:57 sri i just can't track down this stupid bug
12:57 judofyr purl: puh, boehm isn't *the* GC. you silly bot.
12:57 purl judofyr: i'm not following you...
12:57 marcus been bugging p5p, but they think we're just stupid for creating circular references.
12:57 sri t/mojolicious/websocket_lite_app.t is now broken on master + Perl 5.10
12:58 sri 5.10, 5.12, 514… you may only pick two!
12:58 judofyr ah, who needs circular references anyway? it's not like we have closures that closes over *everything* inside them either…
12:59 sri yea, those damn closures are the real problem
13:00 judofyr well, I'd say closures + no proper GC = the real problem
13:03 sri i pretty much know that the bug is caused by some object being cleaned up too early
13:03 sri and causing a cleanup chain reaction
13:03 judofyr s/weaken//
13:04 sri but i'm unable to locate it :S
13:04 judofyr ack weaken
13:04 * judofyr states the obvious
13:37 SSergey joined #mojo
13:47 GitHub79 joined #mojo
13:47 GitHub79 [mojo] kraih pushed 1 new commit to master: http://git.io/cKghHQ
13:47 GitHub79 [mojo/master] fixed a few more portability issues - Sebastian Riedel
13:47 GitHub79 left #mojo
13:47 sri this is just some small stuff i found on the way
13:48 sri not sure how much more time i can invest into the hunt, we might have to leave the 5.12 compatibility issue open for a while
13:49 sri pretty sure it's some weaken/cleanup timing issue
13:52 SSergey for my 5.10.1 all tests now passed, and fail for 5.12.3 (windows), holds on t/mojolicious/websocket_lite_app.t ......... 1/78
13:52 sri i know
13:52 SSergey oh ok
13:53 sri i reverted the 5.12 "fix"
13:53 sri t just shifts the problem but doesn't fix it
13:53 SSergey clear
13:55 sri hopefully more people will join the bug hunt
13:55 sri i'm running out of ideas
13:56 sri at least the test is failing consistently on 5.12 now
13:57 SSergey Sorry i cant help with this, my level is too low
13:57 sri btw. t/mojolicious/embedded_lite_app.t fails randomly too, prolly same cause
13:58 sri just 50% of the time though
14:02 judofyr I love those 50% bugs
14:02 sri they are not as bad as those 1% bugs ;p
14:04 * sri goes back to bashing his head against the wall
14:07 elb0w sri if you didnt know either would you learn a lisp or ruby
14:07 elb0w i think ive narrowed it down to what I want to dive into next
14:10 judofyr elb0w: there's a huge difference between lisp and lisp
14:10 judofyr Common Lisp vs Scheme vs Clojure
14:10 elb0w yeah sorry
14:10 elb0w should of said Clojure
14:10 judofyr oh
14:10 judofyr elb0w: Clojure is more different than Perl
14:10 judofyr you'd probably learn more new stuff
14:11 elb0w But is it stuff that would expand how I looked at other languages like perl?
14:11 elb0w or just viable in lisp like languages
14:13 andrefs joined #mojo
14:20 KayT3 joined #mojo
14:24 metaperl joined #mojo
14:25 metaperl|2 joined #mojo
14:31 metaperl|2 hmm
15:20 thowe joined #mojo
15:21 thowe Hi
15:21 purl privet, thowe.
15:51 kaare joined #mojo
16:04 ispy_ joined #mojo
16:24 tabbi joined #mojo
16:28 yko joined #mojo
16:34 metaperl joined #mojo
16:34 Foxcool joined #mojo
16:34 Foxcool_ joined #mojo
16:40 baton8 joined #mojo
16:41 sri OH NOES
16:41 sri and i found the bug
16:41 sri i'm so stupid
16:46 sri bet this caused many random errors in the past
16:47 grim_fandango joined #mojo
16:47 grim_fandango joined #mojo
17:01 crab what is it?
17:01 purl it's it!
17:02 sri the id generation code
17:03 sri so far i based it on the hex value of the stringified hash
17:11 GitHub42 joined #mojo
17:11 GitHub42 [mojo] kraih pushed 1 new commit to master: http://git.io/5uOMxw
17:11 GitHub42 [mojo/master] fixed more portability issues - Sebastian Riedel
17:11 GitHub42 left #mojo
17:12 sri under the right circumstances a new connection could get an id that was still in use because the hash was destroyed earlier than the connection got cleaned up
17:12 sri which led to both connections being dropped
17:13 noganex joined #mojo
17:14 sri no such thing as a cheap unique id :S
17:18 KayT3 joined #mojo
17:23 andrefs joined #mojo
17:24 sri tempire: http://railscasts.com/pro # :o
17:25 sri it's a cool idea
17:29 ispy_ joined #mojo
17:36 crab hmm. i wonder if that could have had anything to do with the TLS problems.
17:38 mateu joined #mojo
17:39 sri crab: what tls problem?
17:40 sri *+s
17:41 crab when i used ./myapp daemon --listen https://blah, i would get randomly closed connections every once in a while.
17:42 crab it was long ago, once i started using nginx for tls it stopped happening. but now that i put it that way, it doesn't sound like this.
17:42 sri it's unlikely you'd encounter that in a real world setting
17:42 sri IO::Socket::SSL spews a lot of random errors
17:44 crab yeah.
17:45 sri it's still the best ssl module we have though
17:46 MojoGuest698 joined #mojo
17:46 MojoGuest698 From: http://news.ycombinator.com/item?id=1277067 (3 hits)
17:46 MojoGuest698 left #mojo
17:47 MojoGuest698 joined #mojo
17:47 MojoGuest698 From: http://news.ycombinator.com/item?id=1277067 (4 hits)
17:47 MojoGuest698 left #mojo
17:57 GitHub169 joined #mojo
17:57 GitHub169 [mojo] kraih pushed 1 new commit to master: http://git.io/TZ3P6w
17:57 GitHub169 [mojo/master] improved unique id generation slightly - Sebastian Riedel
17:57 GitHub169 left #mojo
18:02 tempire I wonder how long it takes ryan bates to do a railscast
18:03 tempire given that they're not as polished
18:04 tempire I'd be willing to do a pro if people would pay that amount for it.  I don't think the perl community could/would support that, though.
18:06 sri yea, would be rather hard in the perl community
18:06 sri but i could imagine learning perl screencasts would sell
18:08 tempire is there a market for that?  The pro topics would have to be more about generic perl anyway.
18:08 sri i wouldn't be surprised, there's a perl programmer shortage, perl is still popular in many big companies
18:09 sri many surely wouldn't mind paying for subscriptions to get beginners to intermediate
18:09 tempire hmm
18:10 tempire I guess the question becomes, "what is intermediate"
18:10 sri i suppose it would require some aggressive advertising though to reach those companies
18:10 tempire and what is advanced, for that matter
18:10 sri the 3 perl books
18:11 tempire ah.  learning, programming, and mastering.
18:11 sri learning, intermediate, mastering
18:11 sri programming stands on its own as a reference imo
18:11 sri (at least once the next edition is out)
18:12 tempire that's a lot of initial work for no promise of a return.
18:12 * tempire sighs
18:12 sri not saying the chances of success are high ;p
18:13 tempire I'm disappointed.
18:14 tempire the rails casts about page touts that many episodes have been converted into ascii.
18:14 tempire I thought it was going to be ascii art video.
18:14 sri lol
18:14 tempire but it's just typed out tutorials
18:14 tempire boooo
18:37 bduggan joined #mojo
18:43 baton8 joined #mojo
18:47 ispy_ joined #mojo
19:12 tempire wow
19:13 tempire just say the season finale of alphas
19:13 tempire *saw
19:13 tempire freaking awesome
19:32 GitHub107 joined #mojo
19:32 GitHub107 [mojo] kraih pushed 1 new commit to master: http://git.io/yaB-Ug
19:32 GitHub107 [mojo/master] added a few more command tests - Sebastian Riedel
19:32 GitHub107 left #mojo
19:33 * sri agrees with tempire
19:35 * tempire moves on to supernatural season 7
19:35 sri not sure why but i kinda like terra nova
19:35 * sri has not watched supernatural since season 3 or so
19:36 tempire it become *really* good about season 5
19:36 tempire when they started to deal with the apocalypse
19:41 alnewkirk joined #mojo
19:50 MojoGuest894 joined #mojo
19:50 MojoGuest894 From: http://www.google.com/url?sa=t&source=web​&cd=11&ved=0CCIQFjAAOAo&url=http%​3A%2F%2Fdev.xantus.org%2F&rct=j&q=web​socket%20demo&ei=2RGKTsTNIrCrsAKYsfjKDw&a​mp;usg=AFQjCNGw6nZv0wO42uzUnHhWcD0vcKNULg&amp​;sig2=mIuh28iNJKHAV_NRa8F6PA&cad=rja (1 hits)
20:07 Eugene joined #mojo
20:33 perlrocks Twitter: "Perhaps it is about time # MoJo learn Python and Perl can stop with this farce. # # Seriously linenoiseinducingcrapfest" (nl) --Simon_Gee http://twitter.com/Simon_Ge​e/status/120959590958514176
20:42 abra joined #mojo
21:01 perlrocks Twitter: "@ Simon_Gee I grab even back to Perl because I write for years, ditto MoJo. WMI is also not hugely out I think. [1 / 2]" (nl) --t1mdp http://twitter.com/t1mdp/status/120965762377711616
21:19 smpb joined #mojo
21:21 grim_fandango joined #mojo
21:22 s0va joined #mojo
21:22 s0va hello :)
21:23 s0va sri: i think that there are some weird issues with ioloop :)
21:23 ispy_ joined #mojo
21:23 s0va it tries to read from handles marked for reading and tries to write to handles marked only for reading
21:23 s0va this affects EV implementation.
21:24 sri s0va: feel free to send a test case
21:35 s0va sri: how can i submit test case?
21:36 s0va fork your repo, add the test case and issue pull request?
21:36 sri that's one way, or you could first nopaste the patch to discuss it with me
21:36 s0va oh, seems reasonable
21:37 sri you seem to be doing rather crazy things, therefore discussing first might be a good idea
21:45 s0va ok, this one yells like crazy that method connected cannot be located in package io::file (client:149)
21:45 s0va http://pastebin.com/cRXCUQsC
21:45 s0va well... that's reasonable :)
21:45 sri s0va: also, i think you're mistaking Mojo::IOLoop for a general purpose event loop
21:45 sri it is not
21:45 s0va i know :)
21:47 sri https://gist.github.com/1239684 # here's a full featured AnyEvent backend
21:48 s0va why don't want to include it into mojolicious?
21:48 s0va you included EV :)
21:48 sri because i don't like AnyEvent
21:48 sri and i don't want to maintain the backend
21:49 s0va btw, while i tried to fix mojox::run
21:49 sri libev on the other hand is great, i don't care too much for the Perl api, but it gets the job done
21:49 s0va i discovered that EV kidnaps SIGCHLD handler
21:50 s0va just FYI :)
21:50 sri now that's between you and EV
21:50 s0va i replaced SIG{CHLD} with EV::child and it works
21:51 s0va i don't know about hypnotoad but i think that it's using sigchld handler too...
21:51 s0va but i guess that it is working...
21:52 sri if something like that broke in hypnotoad i would be flooded with reports
21:52 s0va do you know any other project based on ioloop?
21:53 s0va i'm suggesting you to port everything to EV :>
21:53 sri nope, and i hope there aren't many
21:53 sri what do you mean by "everything"?
21:53 s0va mojo::server::daemon and mojo::useragent
21:53 sri umm
21:53 sri they already use it
21:54 sri everything in mojolicious uses EV if it's available
21:54 sri or AnyEvent if the backend is loaded
21:54 s0va i'm suggesting porting client/server implementations directly to EV
21:54 s0va :>
21:54 sri that makes no sense
21:54 purl Chewbacca!
21:54 s0va and dropping ioloop completely :)
21:55 sri ignoring the C compiler requirement, how would i implement the blocking APIs?
21:55 s0va oh, goddamn blocking apis :)
21:56 s0va Coro has the answer... but this answer sux really bad.
21:56 sri the whole point of Mojo::IOLoop is for us to be a) event loop agnostic, lets face it they are like religions and b) to have mixed blocking/non-blocking APIs
21:57 sri i don't think Coro is there yet
21:58 s0va blocking apis are provided by AnyEvent, ehehehe
21:58 s0va https://metacpan.org/module/Coro::AnyEvent
21:58 s0va check Coro::AnyEvent::sleep
21:59 * sri still hopes that some day we will have working native threads in perl
22:00 s0va in perl8? :)))
22:00 s0va coro works suprisingly well :)
22:00 s0va https://metacpan.org/module/Coro::Socket
22:00 s0va that's just insane :)
22:01 sri i've worked with Coro in the past
22:11 GitHub86 joined #mojo
22:11 GitHub86 [mojo] kraih pushed 1 new commit to master: http://git.io/PSU9FQ
22:11 GitHub86 [mojo/master] removed obsolete portability workaround - Sebastian Riedel
22:11 GitHub86 left #mojo
22:23 mspo joined #mojo
22:30 andrefs joined #mojo
22:40 metaperl joined #mojo
23:18 mateu joined #mojo
23:43 alnewkirk joined #mojo

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