Camelia, the Perl 6 bug

IRC log for #mojo, 2011-05-07

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

All times shown according to UTC.

Time Nick Message
00:15 vel left #mojo
00:16 mattastrophe left #mojo
01:00 Akron sri: Not very sophisticated and in many points too complicated, but here you are: https://github.com/Akron/mojo/bl​ob/decouple-SAX/lib/Mojo/SAX.pm
01:00 Akron A passable object like "$current" for tree construction is still missing. Resume does not work etc.
01:01 Akron Good night!
01:01 Akron left #mojo
01:03 cyril left #mojo
01:13 jpablobr joined #mojo
01:20 xaka left #mojo
01:31 j3nnn1 joined #mojo
02:03 fmerges left #mojo
02:08 DaTa left #mojo
02:08 DaTa joined #mojo
03:16 xaka joined #mojo
03:26 will joined #mojo
03:27 PerlNinja joined #mojo
03:27 PerlNinja o/ evening/morning/whatever piece of day it is
03:31 miyagawa PerlNinja: got dotcloud invite yet?
03:32 PerlNinja miyagawa: not yet, but I'm not in a rush :) Just had some ideas I wanted to experiment with but since I'm up to my eyeballs in work-that-pays-bills I don't have much time for experimentation
03:34 PerlNinja but if you got one burning in your pocket it'd be welcome :D
03:34 miyagawa already sent you in a dm
03:34 miyagawa :)
03:40 PerlNinja yah just saw it :) account created, i'll go experiment later on :D gotta work first before bossman murders me hehe
03:41 miyagawa cool
04:00 will PerlNinja, hope you have better luck than me.. after an hour I'm still waiting for my dotnet service to "boot"
04:01 PerlNinja will, well, poke miyagawa then :) I'm pretty sure he'd want to hear about things like this
04:01 will s/net/cloud/
04:04 will not a big deal really.. I sent an email off to support.. we'll see what happens
04:07 PerlNinja should be solved, I'm just thinking they're busy as all hell with all us crazy perl people experimenting :D
04:20 xaka|2 joined #mojo
04:20 xaka left #mojo
04:23 miyagawa will: try destroy and re-deploy
04:24 will I'll give it a shot
04:27 will still getting, error: The service is still booting. It will be ready in a few minutes.
04:30 miyagawa yeah confirmed
04:38 xaka joined #mojo
04:38 xaka|2 left #mojo
04:48 Foxcool joined #mojo
04:59 Foxcool left #mojo
05:04 perlrocks Twitter: "Mojolicious-Plugin-ContentManagement 0.012 by MEMOWE - http://frepan.org/~memowe/Mojolicio​us-Plugin-ContentManagement-0.012/" --cpan_new http://twitter.com/cpan_ne​w/status/66730182882492416
05:38 ilbot2 joined #mojo
05:38 Topic for #mojo is now join the revolution ♥ http://mojolicio.ushttp://irclog.perlgeek.de/mojo/today
05:39 Foxcool left #mojo
05:44 kaare joined #mojo
05:46 Foxcool joined #mojo
05:58 miyagawa will: it should be fixed now. try now
06:01 Foxcool left #mojo
06:03 Trelane joined #mojo
06:13 Foxcool joined #mojo
06:16 ka2u joined #mojo
06:19 Foxcool left #mojo
06:58 Foxcool joined #mojo
07:06 sri http://blogs.perl.org/users/opossumpetya/​2011/05/mojo-ppl-what-have-you-done.html # d'oh!
07:10 PerlNinja *snort*
07:10 PerlNinja reads as: "The cheap host that I am using has perl 5.8 and I don't want to go to a host that does have it because it will cost me 2 whole dollars more"
07:14 abra left #mojo
07:15 Foxcool left #mojo
07:17 Foxcool joined #mojo
07:26 abra joined #mojo
07:35 Foxcool left #mojo
07:45 xaka left #mojo
07:46 Foxcool joined #mojo
07:55 xaka joined #mojo
08:06 Foxcool left #mojo
08:06 xaka left #mojo
08:10 Kulag left #mojo
08:10 Kulag joined #mojo
08:25 Foxcool joined #mojo
08:26 AigleNoir left #mojo
08:27 AigleNoir joined #mojo
08:31 perlite_ joined #mojo
08:34 gshank left #mojo
08:34 gshank joined #mojo
08:34 perlite left #mojo
08:34 perlite_ is now known as perlite
08:35 will left #mojo
08:35 yko_ left #mojo
08:35 yko joined #mojo
08:38 cyril joined #mojo
08:40 sri hmm... i don't like that Mojo::SAX :/
08:49 * PerlNinja misread that as "I don't like that Mojo::SEX" ... :/
08:51 PerlNinja *throws himself into the alligator pit*
08:54 yko hey hey, there's not enough space for two persons!
08:54 * yko throws PerlNinja out!
08:55 Foxcool left #mojo
08:57 sri heh
08:57 PerlNinja aww :(
09:00 PerlNinja ok time to relocate the office ... from one starbucks to another :P hehe
09:00 PerlNinja left #mojo
09:04 jwang left #mojo
09:04 jwang joined #mojo
09:10 Foxcool joined #mojo
09:10 jpablobr left #mojo
09:11 non joined #mojo
09:13 hidek left #mojo
09:13 hidek joined #mojo
09:28 Foxcool left #mojo
09:40 perlrocks Twitter: "Wish there were more big projects deprecating #perl 5.8 support, then we wouldn't have these discussions. http://t.co/tiXuRp5 #mojolicious" --kraih http://twitter.com/kraih/status/66799653152227328
09:46 cyril left #mojo
09:47 sri maybe i need to blog about it
10:17 sri or maybe not
10:48 Foxcool joined #mojo
11:08 Foxcool left #mojo
11:12 Foxcool joined #mojo
11:21 cyril joined #mojo
11:28 Foxcool left #mojo
11:40 Foxcool joined #mojo
11:47 cyril left #mojo
11:47 cyril joined #mojo
11:47 Foxcool left #mojo
11:54 cyril left #mojo
12:00 cyril joined #mojo
12:08 Akron joined #mojo
12:08 Akron sri: The namespace thing makes it more complicated than needed maybe. Without that it's pretty simple.
12:16 sri Akron: yea, i don't like that at all
12:16 Akron Namespaces at all?
12:17 Akron ;)
12:17 sri same for cut n pasted entity code
12:17 sri and of course the fundamental flaw
12:18 Akron Yes - that's true. That was a workaround as Mojo::Util does not provide that.
12:18 sri it is not stream based
12:18 Akron You mean the document. Any suggestions on that?
12:19 sri well, like that it is pretty useless
12:20 Akron Hm. It's still useful for things like OpenGraph protocol etc. Have you any ideas how to make it more stream based?
12:21 sri i thought you knew, streams are kinda the whole point of sax ;p
12:23 sri anyway, the whole namespace code alone scares me a lot
12:23 Akron It's not "stream api for xml". :P
12:23 sri since we don't even need it for dom
12:23 Akron Yes - okay.
12:24 Akron But any suggestions for make it more stream based?
12:24 sri i've not thought about that yet
12:25 Akron Okay. I do hate this position thing as well. :)
12:26 sri ok, basically what i was expecting is the current dom module split into a stream based sax parser and a separate dom module, without any new features
12:28 Akron The skip feature is needed for script/style etc.
12:29 memowe_ is now known as memowe
12:32 perlrocks Twitter: "@ @ RT 2mad4milk kraih: Released mojolicious # 1.3 (Tropical drink), enjoy! :) # Perl http://toastlet.iamtoast.biz/tw82884257" (de) --toastlet http://twitter.com/toastle​t/status/66842802360811521
12:34 will joined #mojo
12:35 j3nnn1 left #mojo
12:52 will left #mojo
12:55 will joined #mojo
12:59 will_ joined #mojo
12:59 will left #mojo
12:59 will_ is now known as will
13:23 varta_ left #mojo
13:38 Foxcool joined #mojo
13:44 Foxcool left #mojo
13:52 Foxcool joined #mojo
13:59 Foxcool left #mojo
14:02 wander joined #mojo
14:02 wander left #mojo
14:52 snearch joined #mojo
15:00 Akron left #mojo
15:03 mattastrophe joined #mojo
15:08 marcus I agree with sri
15:08 marcus SAX without streaming is kinda pointless
15:40 will left #mojo
15:53 bosphorus joined #mojo
15:58 will joined #mojo
16:07 will left #mojo
16:26 Akron joined #mojo
16:27 Akron left #mojo
16:28 cyril left #mojo
16:28 cyril joined #mojo
16:56 xaka joined #mojo
17:02 Foxcool joined #mojo
17:13 cyril left #mojo
17:23 cyril joined #mojo
17:39 Foxcool left #mojo
19:09 sri DaTa: suppose you didn't find the freebsd bug? :/
19:12 DaTa sri: no :)
19:12 sri :,(
19:15 sri hmm, could it be a 64bit problem?
19:15 sri the problem would be in Mojo::Transaction::WebSocket then
19:16 DaTa uname -m says amd64, should i try to find a 32bit freebsd box?
19:17 cyril left #mojo
19:17 sri if the bug appears on your box then it's not that
19:17 sri your perl is compiled with 64bit support?
19:17 DaTa use64bitint=define, use64bitall=define
19:18 sri ok, not the bug then
19:18 sri it's only the 64bit test case failing though?
19:19 sri https://github.com/kraih/mojo/blob/master​/t/mojolicious/websocket_lite_app.t#L461
19:21 DaTa no, the test above fails too, but not every time, only sometimes
19:22 sri ok, something is very very wrong there
19:25 sri can you run "MOJO_USERAGENT_DEBUG=1 ./script/mojo test t/mojolicious/websocket_lite_app.t" and nopaste the result?
19:25 sri ideally of a passing and failing one
19:25 DaTa hmm.. can't reproduce that anymore, yesterday 'hi!' test failed in very strange way sometimes, got: was hihihihi..., and expected hi!hi!hi!...
19:26 sri O
19:26 sri o
19:28 sri http://www.cpantesters.org/cpan/report​/50d7ee06-77a7-11e0-9448-31624f584266 # this result i suppose
19:34 DaTa sri: exactly
19:36 DaTa sri: http://data.rbfh.de/mojo-test.tar.xz i've repeated the the command line above a couple of times
19:37 sri just nopaste please
19:37 DaTa sri: there are non printable characters in it
19:37 sri don't really care about those
19:38 non left #mojo
19:38 DaTa don't know how to paste 800k of output :)
19:39 sri :S
19:39 sugar joined #mojo
19:39 sri i just want to know if it used a fresh connection or accidentally kept it alive between broken sessions :/
19:41 DaTa http://data.rbfh.de/mojo-test.97.txt
19:41 sri bluescreen, RLY?
19:41 DaTa oh, s/97.//
19:44 DaTa *g*
19:46 sri hmm
19:48 xaka left #mojo
19:49 sri the first failing test is interesting, the full http message never arrives
19:51 sri DaTa: could you do the same with MOJO_DAEMON_DEBUG=1 instead?
19:51 sri i think the bug could be somewhere around https://github.com/kraih/mojo/blo​b/master/lib/Mojo/IOLoop.pm#L1279
19:54 sri MOJO_IOLOOP_DEBUG=1 might be a good idea too
19:55 DaTa both separate or all in one?
19:55 sri both in one should be fine
19:57 sri i'm starting to see a pattern, but it makes no sense
19:58 DaTa http://data.rbfh.de/mojo-test.txt
20:00 cyril joined #mojo
20:02 sri i don't get it
20:05 sri those are fresh tcp connections
20:05 sri yet they somehow receive data from a previously closed one
20:06 sri and it is somehow freebsd specific
20:06 * sri doesn't get it
20:07 sri i don't even see anything that makes those tests special
20:07 DaTa i guess it's a freebsd bug
20:08 sri it makes no sense
20:10 sri DaTa: could you try one last thing?
20:10 sri https://github.com/kraih/mojo/blo​b/master/lib/Mojo/IOLoop.pm#L808
20:10 sri change that line to "unless (1) {"
20:10 purl sri: that doesn't look right
20:10 sri purl: shut up!
20:10 purl Yes, sir.
20:11 * sri pats purl
20:11 * purl bites!
20:11 sri :o
20:13 DaTa *g*
20:13 DaTa ok, in the meantime i've catched a hihi vs hi!hi! thing: mojo-test.19.txt
20:14 sri those are not so important anymore, we know it gets data from a previous connection somehow
20:15 sri just another case of the same problem
20:16 sri a tainted buffer would explain it, but that couldn't be freebsd specific...
20:16 DaTa mojo-test.windows.txt
20:18 sri ok no difference
20:18 sri thanks
20:18 sri i'm out of ideas now
20:21 sri if i had a freebsd box i would check one buffer after another now
20:26 sri the remainging half of the 500 page should be rather easy to track down
20:26 sri oh i have an idea
20:27 sri DaTa: can you dump $ua at https://github.com/kraih/mojo/blob/master​/t/mojolicious/websocket_lite_app.t#L413
20:28 sri if the remaining half is not in there we know it is something on the socket layer
20:33 jwang left #mojo
20:37 sri guess shortterm i'll just disable the unit test on freebsd -.-
20:42 snearch left #mojo
20:49 DaTa sri: mojo-test.dump.txt
20:52 sri ok, now we are on to something
21:00 sri for some reason there is a leftover transaction associated to a connection
21:01 sri but i'm running out of options again :/
21:03 sri yea, no point for me to continue
21:11 GitHub57 joined #mojo
21:11 GitHub57 mojo: master Sebastian Riedel * 7bcd9a6 (1 files in 1 dirs): disable WebSocket tests on FreeBSD for now - http://bit.ly/kNYIbD
21:11 GitHub57 left #mojo
21:11 sri the pragmatic solution
21:17 sugar_ joined #mojo
21:19 sugar left #mojo
21:25 GitHub182 joined #mojo
21:25 GitHub182 mojo: master Sebastian Riedel * ea3d559 (2 files in 2 dirs): made non-blocking read/write slightly more resilient - http://bit.ly/iL0kGF
21:25 GitHub182 left #mojo
21:31 DaTa sri: would access to freebsd help? i can setup a vm
21:41 sri yes it would
22:00 j3nnn1 joined #mojo
22:08 will joined #mojo
22:19 cyril left #mojo
22:23 kaare left #mojo
22:24 will left #mojo
22:54 perlrocks Twitter: "mojo - Mojolicious: Perl Duct Tape For The HTML5 Web! http://bit.ly/gvMUtq" --github_pl http://twitter.com/github_p​l/status/66999272289214464
23:24 sugar_ left #mojo
23:52 DaTa narf, installed a fresh freebsd and the tests work fine there :)

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