Camelia, the Perl 6 bug

IRC log for #mojo, 2012-06-09

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

All times shown according to UTC.

Time Nick Message
00:08 preflex_ joined #mojo
00:27 ispy_ joined #mojo
00:30 vervain joined #mojo
01:50 mire joined #mojo
01:59 rem_lex|pivo joined #mojo
02:10 mire joined #mojo
02:38 ispy_ joined #mojo
02:58 balazs joined #mojo
03:09 Leandr joined #mojo
03:15 * sri thinks the code name for 3.0 will be Top Hat
03:34 sri naah, it will be Rainbow
03:34 sri U+1F308
03:48 russum joined #mojo
03:52 tempire rainbow top hat
03:52 russum Hi, is the code that is located after shift->render(…) suppose to execute or not, i.e. does the render work like "return" which does not execute any code after it?
03:52 tempire render does not return
03:57 russum that's weird because I'm pretty sure it worked that way for me in the past… so do I simply call return; after render if I don't want the code that goes after to execute?
04:00 tempire that would do it
04:01 russum ok, thanks!
04:07 ispy_ joined #mojo
04:07 russum Hey, if you don't mind, could you take a look at here - http://mojocasts.com/e3 at 0m50s it shows that there are two renders go one after another, but the second one never executes (or only if the first one didn't)
04:08 russum or does this apply only to actual code and not to another render, I'm kinda confused..
04:16 Leandr joined #mojo
04:40 tempire where's the confusion?
04:40 tempire oh, I see what you're saying.
04:41 tempire the second render does happen, but it doesn't overwrite the first render
04:41 tempire it would be more clear in that example to return, but it doesn't matter.
04:51 russum ok, I see, I assumed because of that example that render does return too, thanks for clearing this up for me!
05:11 ovnimancer joined #mojo
05:29 xaka joined #mojo
05:42 * sri accidentally a blog http://blog.kraih.com/mojolici​ous-in-the-cloud-hello-heroku
05:42 Foxcool joined #mojo
06:45 mandreacchio joined #mojo
06:49 Vandal joined #mojo
07:00 cosimo joined #mojo
07:48 AmeliePoulain joined #mojo
08:19 Grauwolf joined #mojo
08:34 black joined #mojo
08:46 mire joined #mojo
08:54 cosimo joined #mojo
09:01 ladnaV joined #mojo
09:19 AmeliePoulain How to user the method from controller from another (non-mojo script)?
09:40 black AmeliePoulain: I used to include the Module in lib folder, then `use Module`, then call the method.
10:02 alexeiras joined #mojo
10:24 Foxcool joined #mojo
10:32 snailfromhu joined #mojo
10:50 i_am_batman joined #mojo
10:56 ladnaV joined #mojo
11:57 baton8 joined #mojo
11:58 i_am_batman joined #mojo
12:22 Leandr joined #mojo
12:29 Vandal joined #mojo
13:02 black joined #mojo
13:08 Psyche^ joined #mojo
13:18 Leandr joined #mojo
13:36 Andrei666 joined #mojo
13:37 Andrei666 hello, can I place objects in session?
13:39 Leandr joined #mojo
13:54 Andrei666 how can I send objects from one action to another ?
14:16 ryanc joined #mojo
14:23 i_am_batman joined #mojo
14:24 gryphon joined #mojo
14:46 black joined #mojo
14:47 inokenty joined #mojo
15:14 inokenty joined #mojo
15:44 jnap joined #mojo
15:47 Andrei666 joined #mojo
15:51 i_am_batman joined #mojo
16:02 snailfromhu joined #mojo
16:17 ladnaV joined #mojo
16:25 balazs joined #mojo
16:35 ladnaV joined #mojo
16:40 ispy_ joined #mojo
17:26 GitHub29 joined #mojo
17:26 GitHub29 [mojo] kraih pushed 1 new commit to master: http://git.io/uXrTdQ
17:26 GitHub29 [mojo/master] added rearranging routes recipe to routing guide - Sebastian Riedel
17:26 GitHub29 left #mojo
17:31 sri thought that should be mentioned in the guides
17:32 Foxcool joined #mojo
17:43 alexeiras joined #mojo
18:45 tempire I had forgotten that you could do that
18:47 xaka joined #mojo
18:50 sri argh
18:50 sri a fail on cpantesters, and it's a windows bug
18:50 sri http://www.cpantesters.org/cpan/report​/beddaf36-920a-1014-ba4a-f2c07cde9195
18:52 sri i've got a report about that from a german perl monger, the conclusion was that it must be a broken dll or so, since on a fresh install it worked perfectly again
18:52 sri interesting that this is win xp though, the other one was windows 7
18:57 sri now that i'm looking at it, i actually have an idea :o
18:57 sri could it be a buffering problem? we write 3 bytes only
18:59 sri we disable nagle's algorithm, but if that is broken on the platform...
19:02 sri too bad this can't be replicated and tested
19:02 sri maybe longer test strings would make the test pass
19:19 sri if i had to remove those tests completely we would basically lose the ability to interrupt long polling connections in Mojo::UserAgent as a feature... hmm
19:20 sri hmm
19:20 sri not to mention inactivity_timeout and request_timeout
19:21 tempire that does not sound good
19:22 sri it's pretty bad, but if it's not cross platform what choise do we have?
19:22 sri s/s/c/
19:23 tempire the presumption is that this bug has always existed, but that it only showed up now?
19:23 sri yes
19:24 sri i had the person who reported it first try old versions of mojolicious too
19:24 tempire and it's intermittent, so only two machines have had the problem as far as we know
19:25 sri two machines means it is not just a freak accident imo
19:26 tempire right
19:26 tempire I'm saying that it won't change anything to wait a bit and try and figure out the underlying issue
19:26 sri i don't see that happening
19:26 tempire rather than just cut out a chunk of useful stuff for the super green
19:27 tempire does the dude have any idea what dll is broken?
19:28 sri nope, she didn't really investigate either, just tried multiple versions
19:28 sri chances that this might happen to someone who actually can investigate are almost non existent
19:29 sri heck, i wouldn't know where to look if i could replicate it
19:30 sri i'm not proposing to remove inactivity_timeout btw. just the test, which would only make it not a feature anymore and unsupported
19:33 tempire I just don't like removing things based on one person's experience.
19:33 sri two
19:34 sri one can be ignored, but two is a problem
19:35 tempire I'm not saying ignore it, but still, just two?  I feel like we'd be getting more people saying something in here.
19:35 tempire there are people who use windows in here
19:35 tempire And I've installed it several times on different windows machines without a problem
19:35 tempire especially lately with the heroku command
19:36 sri i've been saying that in the past too... never worked
19:37 tempire I lost you, saying what?
19:37 sri wait for windows users to pop and fix their own problems
19:37 sri *+up
19:38 sri windows problems result in complaints, not patches
19:39 tempire craig was dwimperl, what sort of perl did the other one have?
19:39 sri that's why i'm extremely conservative when it comes to windows these days, if a feature doesn't work there remove it or disable it completely (like hypnotoad) :)
19:39 sri strawberry
19:39 sri multiple versions, all didn't work
19:40 tempire need more data!
19:40 sri good luck
19:41 sri you'll just end up with unhappy windows users not choosing mojolicious because it didn't install
19:42 sri on unix it's different, but nobody can debug windows...
19:50 * sri ANGRY
19:50 * sri SMASH WINDOWS
19:55 sri on the other hand, our tests will forever be broken on MirOS with Perl < 5.16... so perhaps it is time to give up on super green
20:03 sri hmm
20:03 sri https://github.com/kraih/mojo/commit/a9​439d4d277056b44dad00e1e12a2a899709280f
20:04 sri that one test exists for 2 years... weird the problem never popped up
20:06 sri even versions that old failed on the machine in question btw
20:11 tempire I'd bet there's one goober piece of software that's screwing it up.
20:14 sri broken winsock2.dll or so
20:14 tempire how would that even happen?
20:15 sri random installers overwriting winsock2.dll?
20:16 sri perhaps even a virus or trojan :)
20:17 sri maybe some rootkit that breaks the ability to disable nagles algorithm ;p
20:24 tempire ok.
20:24 tempire non-blocking question coming up
20:24 tempire prepare yourself
20:26 tempire https://gist.github.com/2902439
20:26 tempire first example works.  latter example works, and then hangs until timeout.
20:29 tempire what's the diff, yo?
20:37 tempire hmm, seems like when there's html, it waits
20:41 sri tempire: wat?
20:41 sri oh dammit... we lost purl again
20:42 sri marcus: HALP!
20:42 xaka joined #mojo
20:42 sri tempire: what do you mean with it waits?
20:43 tempire hangs until inactivity timeout kicks in
20:43 sri that's what should happen
20:43 tempire because of long-polling, I presume
20:43 sri keep alive connections still have watchers attached
20:44 sri the transactions are long finished
20:44 snailfromhu joined #mojo
20:44 sri it's only keep alive for reuse
20:44 sri ->start only returns when there are no more watchers
20:45 sri if you add {Connection => 'close'} to the requests it will finish faster
21:07 gryphon joined #mojo
21:13 tempire wellst.  that is sensible.
21:13 gryphon joined #mojo
21:14 tempire now I'm an expert on keep-alive
21:19 sri \o/
21:24 tempire I suspect that connect => close and use a new useragent object answers 99% of the non-blocking questions that happen/will happen.
21:25 rem_lex joined #mojo
21:33 rem_lex joined #mojo
21:36 xaka1 joined #mojo
21:49 sri tempire: you could also undef $ua to kill all keep alive connections on demand
21:50 sri better yet, just use Mojo::IOLoop->delay
21:50 sri it does not rely on remaining watcher count to stop
22:08 sri tempire: https://gist.github.com/2902751 # delay based version
23:28 * tempire scrambles
23:42 * sri just discovered that adblock can also block like and +1 buttons o/

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