The web in a box - a next generation web framework for the Perl programming language

IRC log for #mojo, 2014-07-06

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

All times shown according to UTC.

Time Nick Message
00:29 gatitskiy joined #mojo
00:49 good_news_everyon joined #mojo
00:49 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/EB8joA
00:49 good_news_everyon mojo/master f0b87c1 Sebastian Riedel: use a different news site in example
00:49 good_news_everyon left #mojo
00:53 klapperl joined #mojo
01:32 d4rkie joined #mojo
01:48 Adura joined #mojo
01:49 klapperl_ joined #mojo
01:53 jberger are there windows users here?
01:53 jberger I'm curious if forkcall works under cygwin
01:53 bpmedley I'm currently in Windows, unfortunately.
01:53 jberger it seems native windows fails
01:53 jberger bpmedley: could you try the tests for Mojo::IOLoop::ForkCall?
01:53 bpmedley In cygwin?
01:54 jberger if you have it, but on whatever you have would be interesting
01:54 jberger cpantesters is telling me that native windows is failing
01:54 jberger but no reports for cygwin yet
02:00 bpmedley jberger, http://pastie.org/9359732
02:00 davido___ joined #mojo
02:01 Adura You did what sri did with Minion.
02:01 Adura But, that worked on Cygwin...
02:02 jberger bpmedley: thanks
02:02 jberger was that cygwin or native?
02:02 bpmedley Native
02:04 Adura Worked on Cygwin here.
02:05 jberger Adura: thanks
02:05 jberger gods I hate windows
02:05 * jberger wonders what made minion work
02:18 gatitskiy joined #mojo
02:22 KCL_ joined #mojo
02:25 punter joined #mojo
02:26 jberger perhaps I should just skip that test on native windows and warn not to use event-based programming in the child on that platform
02:26 sri jberger: this is why i wanted you to add a real app test
02:27 jberger my real app test works
02:27 sri it's not about ->reset
02:27 jberger https://github.com/jberger/Mojo-IOLoop-ForkCall/blob/master/t/mojolicious.t
02:27 sri which app test do you mean?
02:28 sri that one is broken
02:28 sri exact thing that happened with minion
02:28 jberger oh!
02:28 jberger balls I read it wrong
02:29 sri fork() segfaults
02:29 sri or rather core dumps
02:29 jberger but loops passes
02:29 jberger loops.t (the new one!)
02:29 sri which means forkcall is completely useless on windows
02:29 jberger the old mojolicious.t test passed before I called reset
02:29 sri because it will never work in a mojo app
02:30 sri are you sure?
02:30 purl You still have ALL THREE lifelines left!
02:30 jberger sri: http://matrix.cpantesters.org/?dist=Mojo-IOLoop-ForkCall%200.09
02:30 jberger 0.09 included mojolicious.t
02:30 sri then the information Adura gave me would have been wrong
02:31 jberger wtf, no it didn't!
02:31 * jberger checks version history
02:31 jberger I bet I forgot to add it to MANIFEST
02:31 jberger dammit
02:31 sri :)
02:32 sri :(
02:32 sri :)
02:32 Adura There is a very slim chance a the test will pass, I'd bet.
02:32 jberger #dammitstevan
02:33 Adura The interpreter error is seemingly random.
02:33 sri that's what you get for having MANIFEST in the repo
02:33 jberger sri: a fair point
02:33 jberger perhaps I should stop doing that
02:37 genio joined #mojo
02:37 jberger so why does it fail in an app but not in a script?
02:37 sri good luck finding out!
02:38 btyler woo got Mojo::UA working over a unix socket. now just need to figure out how to hook it in clearly enough to be general purpose
02:38 jberger btyler: where have you been?
02:38 jberger its been a long time!
02:39 btyler just quiet
02:42 jberger sri: so how does minion pass? does it fork in its tests?
02:42 sri i don't fork() inside the app anymore
02:42 sri you're pretty much screwed
02:43 jberger if only I had a windows machine to test on
02:43 sri https://github.com/kraih/minion/commit/bf9d7608ae1b3793aa1e148b990351176d2c2439
02:43 Adura Talk to the Strawberry guys, tell them to figure out what Cygwin Perl does better...
02:43 sri i changed completely how to test jobs
02:47 jberger sri: I already had that commit open :)
02:58 jberger sri: where is it that mojolicious prevents closing filehandles on fork
02:59 jberger I seem to recall that that is in there somewhere
02:59 KCL joined #mojo
03:06 btyler here's where I'm at: https://gist.github.com/kanatohodets/a0e421415a3506308d5e I did a fair bit more digging, and I'm starting to come to the conclusion that there isn't a nice way to hook in somewhere and get the results from Mojo::URL->parse to create the unixsock Peer
03:06 btyler but I'd love to be corrected
03:26 jberger oh I see, its not that, its that new handles aren't opened with FD_CLOEXEC
03:33 jberger bpmedley: what version of windows native perl is it?
03:33 bpmedley jberger, This is perl 5, version 18, subversion 2 (v5.18.2) built for MSWin32-x64-multi-thread
03:33 bpmedley jberger, It's ActiveState perl
03:39 jberger thanks
03:40 cpan_mojo Statocles 0.018 by Doug Bell - http://metacpan.org/release/PREACTION/Statocles-0.018 (depends on Mojolicious)
03:49 dexteruk joined #mojo
03:54 basic6 joined #mojo
04:00 jberger sri: I think I'm ready to call this a perl bug
04:00 jberger I'm going to try to get a virtual windows so that I can run some of my own tests and try to narrow it down
04:08 btyler got it working...but the means was perhaps evil: https://gist.github.com/kanatohodets/a0e421415a3506308d5e monkey_patch-renamed Mojo::UA->start to non_unix_start and installed my check for a unix scheme into UA->start (and passed along to start)
04:11 jberger btyler, I would say that the monkey patching is rather evil, but the same could be accomplished with subclasses rather easily I would think
04:12 btyler that sounds wildly reasonable
04:23 sujithm joined #mojo
05:09 Eke- joined #mojo
05:37 R0b0t1 Hi, is there a recommended database model provider for mojolicious?
05:37 R0b0t1 I was just gonna use DBI and DBIx::DataModel
05:40 Adura joined #mojo
05:59 crab there is no recommended database model provider for mojolicious
05:59 crab use whatever you want
06:05 irq joined #mojo
06:32 batman jberger: do you create the MANIFEST by hand..?
06:53 Phil21 hmm, is there a clever way to convert a Mojo::UserAgent call into a curl CLI?  useful for handing off to non-perl devs
06:53 Phil21 as API examples
06:53 dexteruk joined #mojo
06:59 Vandal joined #mojo
07:03 batman Phil21: alias mua="perl -Mojo -e'print g(shift)->body'"; mua http://mojolicio.us # super duper basic though...
07:04 Phil21 hmm
07:04 crab oh, that's not how i interpreted phil21's question at all
07:05 Phil21 yeah, not quite
07:05 batman i didn't get the question, so i just guessed as good as i could :/
07:05 Phil21 basically looking for it to spit out curl -X "header here" -d "{json}" etc.
07:05 crab you wanted to convert an invocation of m::u into an equivalent curl invocation?
07:05 Phil21 would take a pretty complicated module :)
07:05 batman it's my "trick", that sometimes go wrong :P
07:05 Phil21 yeah
07:05 Phil21 batman: that is pretty neat though
07:05 crab that's horrible. don't do it.
07:06 Phil21 well, I'm testing a remote API w/ Mojo:UA and it's great
07:06 Phil21 now I get to go through and create a bunch of curl examples :/
07:06 Phil21 was trying to be lazy
07:06 Phil21 I recall seeing something similar for LWP and it nominally worked, but didn't have a use at the time
07:06 batman Phil21: you could just wrap the code you have inside scripts that they can execute..?
07:07 crab presumably the people who need this documentation will be using something other than perl
07:07 Phil21 it's not that really, it's more they are coing to code the production stuff in node or whatever fotm
07:07 Phil21 I will try handing off my perl :)
07:07 Phil21 s/coing/going/
07:07 Phil21 creating spec documentation, to hand off to the real developers basically
07:09 Phil21 but man, loving Mojo the more I hack at it
07:10 Phil21 everything is looking like a nail, now that I have this hammer :)
07:11 preaction isn't there some http proxy thing that will make replayable scripts in various languages?
07:12 * batman will try to understand again
07:12 preaction maybe i was thinking of HTTP::Recorder, which is probably not what you want
07:13 preaction why can't they just install perl and mojolicious? they can install everything else, they're "real" developers
07:13 batman Phil21: do you want a cli tool based on Mojo::UA that can do some of the things curl can?
07:13 Phil21 batman: no, I want to build a Mojo:UA request, and then have it output what I could paste into CLI for curl
07:13 preaction batman: no, he wants to give someone else a shell script using curl that does what he already has working via Mojo::UA
07:13 Phil21 sec
07:13 Phil21 I know I found this before, lemme see if I Can magic up some google
07:13 Phil21 (just not for mojo)
07:14 preaction HTTP::Request::AsCurl
07:14 preaction is the LWP one
07:14 Phil21 ah, yes that's it
07:14 Phil21 it's not a big deal, just figured I'd ask before I hacked something close enough
07:15 batman Phil21: oh. now i get it :)
07:15 preaction bonus: if they have to install/use perl/Mojolicious, they might like it too! you're evangelyzing!
07:15 Phil21 :)
07:15 preaction it's an uphill battle, to be sure, but ... yeah i got nothing
07:16 Phil21 yeah, Perl is getting hard to evangelize
07:17 Phil21 but we still have our main CRM/everything-tracker in Perl
07:17 Phil21 and some of the younger devs I've shown the Mojo webcasts too and they like it, still not as cool for them to put on a resume as convincing folks to let them use Node or whatever though :/
07:20 preaction the exodus from node to go is underway, anyway. and then who knows what... the next language probably hasn't been started yet
07:21 irq_ joined #mojo
07:22 Phil21 yeah, it's a bit ridiculous to me - so much energy expended in reinventing wheels
07:22 batman Phil21: https://gist.github.com/jhthorsen/8a56a5c136f1fd06ff2b
07:22 batman again: super basic, but maybe it's something in the direction of what you want...?
07:22 Phil21 huh. wow cool thanks
07:23 Phil21 I can definitely make this work :)
07:23 batman the nice thing is that you can factor out the callback and put it into any "start" event
07:25 batman remember proper escaping... i tried to be clever about the $body variable, but it's probably not good enough...
07:25 batman the shell works in mysterious ways :)
07:25 Phil21 yeah, it doesn't need to be super robust
07:25 Phil21 sec seeing how it works
07:27 btyler what's the easiest way to set up a http server in mojo when I already have a specific socket handle I'd like it to use?
07:28 btyler I wrote a super naive http server using raw socket accept/printing to the connection filehandle, but it struck me that there's probably a much easier way to do this using mojo components
07:29 btyler context: I'm testing Mojo::UserAgent::UnixSocket, and I'd like to have something for it to talk to
07:30 btyler perhaps I can get away just making use of mojo http tools, and leave the silly socket futzing in place
07:30 Phil21 batman: heh, yeah this is perfect
07:31 batman anytime ;)
07:32 Phil21 I see why you're called batman :)
07:32 batman called, you say..? ;)
07:32 Phil21 ha
07:33 batman i'm not saying i am batman. just that you haven't seen me and bruce wayne in the same room...
07:35 btyler maybe a simpler question: I have the UA posting some JSON to this spit-and-glue http server, but the JSON doesn't appear on the server until the inactivity_timeout hits on the UA
07:35 btyler I have autoflush on for the connection on the server side
07:35 Phil21 you also just taught me the ->to_hash thing too
07:35 Phil21 so much to learn
07:36 btyler mojo is a very big box of goodies
07:36 batman Phil21: yeah, it's rather useful. the reason why it's not there by default is that you can repeat the same header... no idea when that happens though :)
07:49 gatitskiy joined #mojo
08:02 gatitski_ joined #mojo
08:34 yusukebe joined #mojo
08:59 dexteruk joined #mojo
09:01 btyler anyone know why $tx->res might end up with valid html in 'buffer' but ->dom is empty? tests are working on my mac, but failing in my linux vagrant :(
09:05 btyler https://gist.github.com/kanatohodets/7a00f84529066ebab522 if someone has a minute
09:06 btyler (also, if there's a nice mojo way to bypass the http-server crap I'm forking, I'm all ears)
09:07 * batman wonder why anyone would do http over unix socket...
09:07 btyler the one I'm bumping into is docker
09:16 btyler ah ha, character encoding
09:19 btyler and by that I mean the content-length header. so strange that the tests work on OSX but fail on linux, but I expect its something silly related to the hour of the day
09:36 basiliscos joined #mojo
10:10 rem_lex| joined #mojo
10:33 mire joined #mojo
10:53 sri btyler: your hack is also missing support for the Host header
10:53 purl okay, sri.
11:09 sh4 joined #mojo
11:21 Negor joined #mojo
11:29 denis_boyun joined #mojo
11:39 meshl joined #mojo
11:55 sugar joined #mojo
11:55 berov joined #mojo
12:02 mire joined #mojo
12:12 jberger Little known fact, but no one has ever seen me and John Cleese in the same room either.
12:12 jberger Just sayin
12:25 dod joined #mojo
12:25 batman hehe
12:27 batman jberger: let me know if you find something odd with Refine
12:40 sri batman: i might be wrong, but doesn't your version of refine defeat the purpose?
12:41 sri ah, no it doesn't
12:46 * batman got curious
12:47 sri oh, 3 mojolicious talks at yapc::asia
12:49 batman cool :)
12:51 mire joined #mojo
13:01 good_news_everyon joined #mojo
13:01 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/MFgaRA
13:01 good_news_everyon mojo/master d252277 Sebastian Riedel: no need to weaken all the time
13:01 good_news_everyon left #mojo
13:20 batman sri: could you tell me why you thought it defeated the purpose? maybe it's an idea for a new feature :)
13:21 sri private/public names confused me, thought there might be a leak
13:21 batman ok
13:21 batman unittest++ # ftw ;)
13:31 jberger perl -Mojo -E "a('/' => sub { fork })->start" get /
13:31 jberger dies on windows
13:32 sri try lower levels... like Mojo::IOLoop->server
13:32 jberger I'm not sure I know how
13:33 sri find out! :)
13:33 jberger I probably should eh
13:33 jberger ye gods I hate the windows command line
13:35 sri although, it is certainly a perl bug.... so there might not even be something you can do, besides reporting it
13:36 sri you can hack examples/microhttpd.pl i guess
13:36 sri for a quick test
13:37 jberger if I can figure it out, I intend to report it
13:37 jberger also, fyi, this exists: https://www.modern.ie/en-us/virtualization-tools#downloads
13:38 jberger microsoft gives away vbox appliances of windows for development purposes
13:39 sri interesting
13:40 sri the fact that they use rar files is kinda hilarious
13:46 jberger they also give instructions about how to defeat the timeout
13:46 jberger at least for a whil
13:46 jberger while
13:46 jberger they know their target audience :-)
13:58 jberger sri: ok so microhttpd.pl works
13:58 jberger then I add 'return unless fork;' in the stream read callback
13:58 jberger and I start getting attempt to free unreferenced scalar from all over the plae
13:58 jberger place
14:00 punter joined #mojo
14:00 jberger sri: what is multi_accept?
14:00 denis_boyun joined #mojo
14:02 jberger there is line 171 in Mojo::IOLoop
14:02 sri an optimization
14:02 jberger and then every other attempt to free is in Mojo::Reactor::Poll or IO::Poll
14:02 purl an optimization is the mother of all fuckups. or "1) it's an IO problem, 2) it's an IO problem, 3) it's an IO problem" or for those who don't have enough hardware or something people often obsess about pointlessly or see Rules of Optimization or "Premature optimization is the root of all evil"
14:03 sri multiple ->accept calls for one readable event
14:03 sri pretty big optimization actually
14:06 jberger setting max_connections to 1, I now get a different error
14:07 jberger use of freed value in iteration at Mojo/Reactor/Poll.pm line 66
14:12 jberger I'm starting to suspect IO::Poll
14:15 jberger this block seems to have unreferenced scalars https://metacpan.org/source/GBARR/IO-1.25/lib/IO/Poll.pm#L79
14:46 jberger sri: I'm considering a Mojo::Reactor::Select
14:59 KCL_ joined #mojo
15:15 btyler sri: yep, added that, here's what it looks like now: https://github.com/kanatohodets/mojo-useragent-unixsocket/blob/master/lib/Mojo/UserAgent/UnixSocket.pm
15:15 btyler thanks for taking a look earlier
15:16 btyler (re host header)
15:19 dod joined #mojo
15:21 btyler and https://github.com/kanatohodets/mojo-useragent-unixsocket/blob/master/t/basic.t is the test with the 'by hand' http server stuff I was wondering how to simplify with mojo
15:24 D4RK-PH0ENiX joined #mojo
15:33 denis_boyun joined #mojo
15:35 sri jberger: no
15:39 jberger sri: I meant that only to test if IO::Poll is the problem
15:39 sri i'm gonna keep an eye on you!
15:39 jberger how else would you test that?
15:40 sri minimal test case
15:40 purl minimal test case is what you do when your forehead starts bleeding or blood pressure medicine for the frustrated programmer
15:41 irq joined #mojo
15:41 sri i wouldn't be surprised if the IO::Poll xs code was broken somehow
15:42 sri maybe not properly thread safe
15:42 sri the IO module had its last release in 2009... and 36 open issues
15:43 jberger I've been reading through them
16:13 btyler any issue with my hopping on the Mojo::UserAgent namespace? I see MojoX, but those tend to be older. about to cpan-ify M::UA::UnixSocket
16:16 sri btyler: what you're doing with the host seems awkward
16:17 sri path as Host header value seems worse than no Host header
16:18 btyler I suppose localhost is more correct
16:18 sri yea
16:25 btyler sri: fixed, thanks
16:27 bpmedley joined #mojo
16:28 jberger sri: the poll xsub is kinda terrifying (assuming I understand what is going on)
16:32 cpan_mojo Mojo-UserAgent-UnixSocket 0.01 by BTYLER - http://metacpan.org/release/BTYLER/Mojo-UserAgent-UnixSocket-0.01
16:38 davido__ joined #mojo
17:01 sri if anyone wants to hack on http/2 support, i've lowered the requirements https://github.com/kraih/mojo/issues/423#issuecomment-48115497
17:02 sri i think it might actually fit into a minor release
17:02 sri just a bunch of new modules... like Mojo::HPack... and lower level hooks for ALPN/NPN negotiation
17:03 sri don't think it even needs a Mojo::Transaction::HTTP2
17:06 rem_lex joined #mojo
17:08 rem_lex|pivo joined #mojo
17:08 berov jberger: Thank you for the "VirtualBox for Linux" link
17:09 berov I have some failing tests, hopefully now will make them pass :)
17:09 * sri wishes virtualbox wasn't oracle...
17:09 berov sri: me too :)
17:10 berov I think it will be good if there is a Mojo Presentation on YAPC::Europ too
17:10 berov this year in Bulgaria
17:10 dod joined #mojo
17:11 rem_lex| joined #mojo
17:11 berov http://act.yapc.eu/ye2014/
17:11 sri deadline is over
17:12 berov I know the organiser
17:12 berov it is soft deadline
17:13 berov but I am not very familiar with the advanced stuff
17:13 berov of Mojo
17:13 berov still an introduction would be ok
17:16 d4rkie joined #mojo
17:32 kyshtynbai joined #mojo
17:40 edestler joined #mojo
17:57 igorsutton joined #mojo
18:15 _eugen joined #mojo
18:22 batman marcus: does Mojolicious::Plugin::RequestBase still work? i don't see any tests that actually check if it works...
18:23 batman sri: should it still work, now that urls are relative instead of absolute?
18:27 sri batman: what do you mean "now"?
18:28 batman sri: https://github.com/jhthorsen/Mojolicious-Plugin-RequestBase/blob/master/t/basic.t#L20
18:28 batman now == a long time ago, just that i'm still stuck in the stoneage
18:29 sri i don't know which change you're referring to
18:29 batman let me see if i can find it *looking*
18:34 sh4 joined #mojo
18:34 batman i can't find the commit, but i think url_for() used to create absolute paths instead of relative..?
18:35 batman hm... i guess it doesn't really matter what domain it creates, since the browser will prepend with the correct hostname+port
18:38 sri you're imagining things
18:38 mire joined #mojo
18:39 batman ok.
18:40 batman maybe it was just redirect_to() that changed
18:51 abra joined #mojo
19:05 meshl joined #mojo
19:08 basic6_ joined #mojo
19:22 punter joined #mojo
19:50 eoin joined #mojo
19:51 sri batman: redirect_to() did change to allow relative Location values, which are now valid in the new HTTP RFC
19:51 sri but both relative and absolute versions are semantically equal
20:00 punter sri, do you have a notifier to track changes in all RFCs that interest you?
20:01 sri nope, i'm sometimes active in the working groups that interest me
20:01 sri http-wg, hybi-wg, apps-wg...
20:01 punter wow ok
20:03 punter apps-wg looks interesting to me too
20:08 batman sri: ok. thanks
20:09 * sri sometimes wants to remove the stringify overload from Mojo::Collection to make ->find(...)->find(...) mistakes more obvious
20:09 sri punter: apps-wg results are rather meh though
20:10 sri at least i don't remember the last time a good RFC came out of it
20:11 sri then again... we've not seen many good RFCs in general recently
20:12 sri 2616 cleanup was pretty decent... but without any real changes allowed there wasn't much to go wrong
20:12 denis_boyun joined #mojo
20:15 sri all you can hope for is flying under the radar long enough that the corporate sock puppets don't have a chance to kill your spec
20:16 punter Anybody here interested in making the biggest, most badass forums site on earth?
20:17 punter with me?
20:17 punter I have ideas
20:17 punter hm ok
20:18 irq_ joined #mojo
20:18 punter (my ideas can bury Discourse)
20:18 lb my ideas can bury facebook, doesn't mean they will ;)
20:19 sri i soviet #mojo sri buries ideas
20:19 punter it's a matter of will and determination
20:20 sri also marketing
20:21 punter Perhaps, yes
20:21 punter So you think that word of mouth is never enough?
20:21 lb and two container loads of luck
20:24 sri it's interesting how the ghost project is evolving, now the website actually hides the open source part https://ghost.org
20:25 lb hides?
20:25 punter i'll try ghost out
20:28 * sri really doesn't like discourse btw.
20:29 sri can't follow anything in discourse discussions... i'm more into the reddit comment format
20:32 sri btw. does anyone have a really really good way for documenting argument/return value types in POD?
20:32 sri like... solid conventions
20:33 sri i've been meaning to change my documentation style for ages... but trying to document methods with anything even remotely resembling types always ends up totally sucking
20:34 edestler joined #mojo
20:55 batman sri: i like how you document things now... probably because i do the same ;)
20:58 denis_boyun joined #mojo
20:59 tempire sri: I've tried several ways, they're all...just ok I think it's just going to be a matter of picking something and going with it.
20:59 tempire s/just ok I think/just ok. I think/
21:13 basiliscos joined #mojo
21:18 jberger sri: I got it
21:19 jberger and you are not going to believe what it is
21:24 disputin joined #mojo
21:24 jberger I'm trying to think what the best way to right a test for this is
21:25 jberger sri: i'm going to send a preliminary pr
21:25 jberger just to show you the necessary change, and then we can discuss testing it ok?
21:32 good_news_everyon joined #mojo
21:32 good_news_everyon [mojo] jberger created fix_fork (+1 new commit): http://git.io/1gG89g
21:32 good_news_everyon mojo/fix_fork d9aaa88 Joel Berger: fix forking while iterating over file handles with events, specifically using $_
21:32 good_news_everyon left #mojo
21:32 * sri phears
21:33 sri ewwww
21:35 sri why?
21:36 jberger even without the autoviv fix, it is the use of $_ that breaks things
21:37 jberger if you remove the my $f just use $_, it breaks windows when forking
21:39 sri so why did you add "next unless exists $self->{io}{fileno $f};"?
21:40 sri looks like you made two unrelated changes that each need to be tested individually
21:40 jberger because calling $self->{io}{fileno $f}{cb} will autoviv $self->{io}{fileno $f} and thus when you check $self->stop unless keys ... well there are keys
21:40 sri so it has nothing to do with the fork fix?
21:41 jberger it was what I was THINKING was going wrong, and I suspect it is still a bug
21:41 jberger but when I backed it out, the problem didn't return
21:41 jberger so I also when back to $_ (with the block form still, not postfix) and it crashed again
21:42 jberger if you would like to handle them one at a time, do $_ first
21:44 good_news_everyon joined #mojo
21:44 good_news_everyon [mojo] kraih pushed 1 new commit to fix_fork: http://git.io/aJEwDQ
21:44 good_news_everyon mojo/fix_fork 6eab5ff Sebastian Riedel: no need to call fileno twice
21:44 good_news_everyon left #mojo
21:44 jberger hmmmmm
21:44 jberger hold off a minute now
21:44 sri well, i'm not happy with the 4 extra lines... but if it has to be done :(
21:44 jberger it fixed my microhttpd test, but I'm still not able to get forkcall to pass
21:45 sri aha
21:45 * jberger keeps working
21:45 sri that proves Adura's point, that the bug is more or less random, prolly affected by small memory management changes
21:58 sri jberger: think i have a test for the second part
21:59 sri yea, confirmed got a test
22:00 jberger cool
22:01 good_news_everyon joined #mojo
22:01 good_news_everyon [mojo] kraih pushed 1 new commit to fix_fork: http://git.io/cEmihg
22:01 good_news_everyon mojo/fix_fork cedeab3 Sebastian Riedel: test reset with active watchers
22:01 good_news_everyon left #mojo
22:01 good_news_everyon joined #mojo
22:01 good_news_everyon [mojo] kraih fast-forwarded master from d252277 to d9aaa88: http://git.io/9kgCNA
22:01 good_news_everyon left #mojo
22:02 sri oops
22:02 sri how the hell did that happen Oo
22:06 good_news_everyon joined #mojo
22:06 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/OuA1Fg
22:06 good_news_everyon mojo/master 0779e1a Sebastian Riedel: Merge pull request #643 from kraih/fix_fork...
22:06 good_news_everyon left #mojo
22:07 edestler joined #mojo
22:08 jberger well, at least that was useful
22:08 jberger still can't get forkcall to work
22:08 jberger easiest demonstration is still:
22:09 jberger perl -Mojo -E "a('/' => sub { fork })->start" get /
22:09 jberger that wouldn't do anything useful if it worked, but it doesn't
22:10 good_news_everyon joined #mojo
22:10 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/sInT7A
22:10 good_news_everyon mojo/master 2eb6f1c Sebastian Riedel: updated Changes file
22:10 good_news_everyon left #mojo
22:10 sri i wonder how the hell that fast-forward happened
22:11 jberger ¯\_(ツ)_/¯
22:12 sri (╯°□°)╯︵ ┻━┻
22:14 jberger the microhttpd death really did fix though
22:14 jberger I'm not so sure that it is random memory management
22:20 sri interesting http://talinashrotriya.wordpress.com/2014/07/04/web-of-trust-initial-thoughts/
22:26 punter joined #mojo
22:29 edestler joined #mojo
22:39 jberger well that's all I can stare at if for todady
22:55 basic6 joined #mojo
23:00 bluescreen joined #mojo
23:13 d4rkie joined #mojo
23:18 punter joined #mojo
23:35 basic6_ joined #mojo
23:50 basic6 joined #mojo

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