00:06:23  * bnoordhuisquit (Read error: Operation timed out)
03:25:35  * Raynosquit (Remote host closed the connection)
03:37:24  * piscisaureus__quit (Quit: ~ Trillian Astra - www.trillian.im ~)
03:46:09  * Raynosjoined
03:55:58  * piscisaureus_joined
04:04:35  * piscisaureus_quit (Ping timeout: 240 seconds)
04:55:52  * dshaw_joined
05:15:02  * dshaw_quit (Ping timeout: 240 seconds)
06:18:47  * ericktquit (Quit: erickt)
07:08:43  * ericktjoined
07:10:44  * ericktquit (Client Quit)
07:26:53  * sh1mmerjoined
08:36:43  <CIA-111>node: koichik master * rb962ff3 / lib/tls.js :
08:36:44  <CIA-111>node: tls: fix test-https-client-reject fails
08:36:44  <CIA-111>node: Fixes #2417. - http://git.io/_SkASA
08:46:25  <CIA-111>node: koichik v0.6 * ra337ac7 / (lib/http.js test/simple/test-http-pipe-fs.js):
08:46:25  <CIA-111>node: http: fix XMLHttpRequest piped in a writable file stream hangs next request
08:46:25  <CIA-111>node: Fixes #2263. - http://git.io/4wTdeQ
08:46:25  <CIA-111>node: koichik v0.6 * re1c043f / (13 files):
08:46:25  <CIA-111>node: docs: fix docs to not suggest variable leaks
08:46:25  <CIA-111>node: Fixes #2106. - http://git.io/ZucbaQ
08:48:04  * travis-cijoined
08:48:04  <travis-ci>[travis-ci] joyent/node#148 (master - b962ff3 : koichik): The build is still failing.
08:48:04  <travis-ci>[travis-ci] Change view : https://github.com/joyent/node/compare/e6b6075...b962ff3
08:48:04  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/node/builds/450521
08:48:04  * travis-cipart
08:48:49  * Raynosquit (Max SendQ exceeded)
08:54:06  * travis-cijoined
08:54:07  <travis-ci>[travis-ci] joyent/node#149 (v0.6 - e1c043f : koichik): The build passed.
08:54:07  <travis-ci>[travis-ci] Change view : https://github.com/joyent/node/compare/70e6f3f...e1c043f
08:54:07  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/node/builds/450546
08:54:07  * travis-cipart
09:09:33  * Raynosjoined
09:13:58  * ericktjoined
09:38:04  * eddybjoined
09:38:25  <eddyb>hi
09:38:31  <eddyb>I dropped the idea of rewriting node.js
09:38:39  <eddyb>and I made a quick wrapper around it
09:38:47  <eddyb>now... how do I build a simple .node module?
10:05:31  * ericktquit (Quit: erickt)
10:16:50  * mikealjoined
10:18:07  * mikealquit (Read error: Connection reset by peer)
10:56:23  <indutny>eddyb: for each node version?
10:57:34  <eddyb>for the latest version
10:57:41  <eddyb>or.. what do you mean?
11:17:19  <indutny>eddyb: latest version is using gyp for building .node files
11:18:08  <eddyb>yepp
11:18:15  <eddyb>how can I use gyp?
11:18:23  <eddyb>I almost got this beast working
11:19:36  <eddyb>it can produce node.js modules out of something like this http://pastebin.com/010FYuZC
11:28:21  <eddyb>indutny: ^^
11:47:27  <eddyb>c'mon :|
11:47:31  <eddyb>is there a guide, anything?
12:05:15  <eddyb>indutny: how do I build module pretty please?
12:15:33  <eddyb>indutny: you lied :|
12:15:36  <eddyb>there's no gyp installed
12:20:26  <indutny>eddyb: sorry man, I was away
12:20:49  <indutny>eddyb: node is going to use gyp since 0.7.x version
12:20:52  <indutny>and it's the latest
12:21:11  <eddyb>gonna use https://github.com/shinuza/node-cc-module-boilerplate/blob/master/wscript for now
12:22:32  <eddyb>and thanks for the heads up :)
12:42:48  * piscisaureus_joined
12:54:19  <indutny>eddyb: :)
14:39:40  * bnoordhuisjoined
14:43:01  * _eddyb_joined
14:43:24  * eddybquit (Read error: Connection reset by peer)
14:43:28  * _eddyb_changed nick to eddyb
15:15:39  * paddybyersjoined
15:57:27  * bnoordhuisquit (Read error: Operation timed out)
16:38:49  * paddybyersquit (Quit: paddybyers)
16:56:21  * AndreasMadsenjoined
16:56:53  <AndreasMadsen>So now I'm having 6 pull request open to Node.js :)
16:57:30  * piscisaureus_quit (Read error: Connection reset by peer)
16:57:38  <txdv>work faster node.js developers! you can enjoy your life in the next life
16:57:46  <AndreasMadsen>mmalecki: I have solved the IPC channel disconnect and fork background problem: https://github.com/joyent/node/pull/2426
16:58:21  <AndreasMadsen>mmalecki: so you don't have to kill any more puppies by doing process.exit(0)
16:59:04  <AndreasMadsen>mmalecki: The function can with little work be implemented manual in node v0.6
16:59:59  <AndreasMadsen>txdv: Anything there need to be done ?
17:00:07  * piscisaureus_joined
17:00:33  <mmalecki>AndreasMadsen: wow, awesome!
17:01:23  <AndreasMadsen>mmalecki: it required about a day of hacking :)
17:02:32  <txdv>me just trolling
17:03:24  <AndreasMadsen>txdv: I thought so, but I'm felling that I can do anything to day so just wanted to ask :)
17:04:31  <mmalecki>AndreasMadsen: man, do you ever sleep :D ?
17:05:56  <AndreasMadsen>mmalecki: for about 2 hour I was fighting a SIGBUS issue related to closing the IPC, I think I fixed it but please tell me if not.
17:06:21  <AndreasMadsen>mmalecki: Well I have christmas holiday so not much.
17:07:43  <mmalecki>AndreasMadsen: ah, so do I :). you should go and enjoy it :)
17:09:14  <AndreasMadsen>mmalecki: Well I have become addicted to contributing to node.js. Just look at all that blue color: http://img820.imageshack.us/img820/484/skrmbillede20111227kl18.png
17:11:28  <mmalecki>AndreasMadsen: oh, that looks like a big win :). keep up the awesome work, sir!
17:11:48  <AndreasMadsen>mmalecki: win ?
17:12:38  <mmalecki>AndreasMadsen: like, victory :)
17:12:57  <AndreasMadsen>Sure
17:19:17  * paddybyersjoined
17:29:54  * paddybyersquit (Quit: paddybyers)
17:37:17  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
17:40:17  * piscisaureus_joined
18:17:14  * paddybyersjoined
18:24:36  * ericktjoined
18:29:16  * paddybyersquit (Quit: paddybyers)
18:56:15  <ryah>piscisaureus_: yo
19:02:09  <ryah>mjr_: any updates? did you try ben's patch?
19:02:43  <mjr_>We rolled out that patch, but we haven't restarted everything with with yet. Probably should do that now.
19:03:09  * mralephjoined
19:04:11  * piscisaureus__joined
19:04:18  <piscisaureus__>ryah: yo
19:05:06  * piscisaureus__changed nick to piscisaureus
19:08:07  <AndreasMadsen>ryah: if you have finished reviewing https://github.com/joyent/node/pull/2388 could you then land it?
19:08:42  * AndreasM_joined
19:09:29  <ryah>piscisaureus: yo - i can't do the call because im not in the office. any updates?
19:09:50  <ryah>piscisaureus: none from my side - i've been christmas-ing the last few days
19:09:56  <piscisaureus>me too
19:10:07  <ryah>k
19:10:15  <piscisaureus>I have decided to split domains work in two parts
19:10:27  <piscisaureus>one internal interface that just exposes handle groups
19:10:31  <ryah>AndreasMadsen: i will have a look today
19:10:33  <piscisaureus>and allows destroying them
19:11:12  <piscisaureus>the real domains api is a little bit tbd, you've seen my api proposal but I feel we need a better solution for the eventEmitter part of it
19:11:13  <AndreasM_>ryah: awesome
19:11:16  <ryah>mraleph: i retracted my retweet from marak
19:11:36  <ryah>although i think the hook.io stuff is cool :)
19:12:03  <mraleph>ryah: yes it is :-) I did not mean to influence your retweet. I am just curious :-)
19:12:23  <ryah>piscisaureus: that sounds like a good idea
19:12:37  <ryah>handle groups will be useful for isolates
19:13:00  <ryah>piscisaureus: yes i saw your api proposal and i thought it looked good
19:13:00  <mraleph>ryah: I hope he does not interpret halting problem as "it's impossible to make program run forever (-> be fault tolerant)"
19:13:13  <ryah>piscisaureus: but i forget the eventEmitter part - do you have a link?
19:13:43  <ryah>mraleph: that's what it sounds like :)
19:14:35  <piscisaureus>ryah: the EventEmitter part is actually kind of a misnomer... But we need a better story for accepting a connection and assigning the accepted connection to a domain
19:14:52  <piscisaureus>ryah: I will think about it, I'll keep you posted
19:14:54  <piscisaureus>I have some ideas
19:15:05  <piscisaureus>just need to explore the problem space
19:15:48  <ryah>we need a way to hand off one handle to another domain
19:16:09  <piscisaureus>I am not sure tbh
19:16:10  <ryah>but - let's not worry about it yet - better to get the handle groups done
19:16:16  <piscisaureus>yeah
19:16:28  <piscisaureus>I'm slow, but I'll get going again now
19:16:38  <piscisaureus>it actually works but it's all in one .h file now
19:17:34  <piscisaureus>ryah: btw how far is the domains work now, I've not really been following it
19:18:51  <ryah>piscisaureus: isolates?
19:18:57  <piscisaureus>er, yeah
19:19:09  <piscisaureus>sorry, I meant isolates
19:19:14  <ryah>they work, kind of
19:19:34  <ryah>you can't kill them - but can run eio stuff in either one
19:19:42  <piscisaureus>is it race-free? do they support child processes?
19:19:43  <ryah>thanks to paddy's patch
19:20:03  <ryah>no
19:20:15  <ryah>but you can load an addon in each
19:20:25  <piscisaureus>ah, that's cool
19:21:01  <ryah>piscisaureus: https://github.com/joyent/node/tree/isolates2/test/addons/shared-buffer
19:21:32  <ryah>next step is ipc pipe
19:21:40  <ryah>which im going to do today
19:22:01  <piscisaureus>cool
19:29:23  <ryah>mraleph: did you look at that next-tick thing?
19:30:09  <ryah>mraleph: i wouldn't be surprised if it was your regression - we didn't touch that code much
19:30:36  <mraleph>ryah: no I did not have a chance.
19:30:46  <mraleph>ryah: but I will try to that next year.
19:32:26  <ryah>mraleph: ok - arlo just added it to our CI http://arlolra.no.de/bench/next-tick-2
19:36:19  <ryah>i wish pull requests had the link to the remote repo in them
19:37:50  * AndreasM_quit (Remote host closed the connection)
19:38:15  <mmalecki>ryah: whatever you're doing with github, use hub: http://defunkt.io/hub/
19:42:18  <ryah>mmalecki: okay - but it doesn't seem to have any PR functionality
19:42:22  <ryah>otherwise looks good
19:42:32  <mmalecki>ryah: it does
19:42:48  <mmalecki>ryah: probably better explained here https://github.com/defunkt/hub#readme
19:43:54  <mmalecki>ryah: it doesn't have the functionality you want, but you can add remotes way easier
19:44:12  <mmalecki>apply patches using pull request links, etc.
19:45:54  <ryah>so whats' the best way to try out https://github.com/joyent/node/pull/2388
19:47:28  <mmalecki>using hub? `hub am https://github.com/joyent/node/pull/2388`, I'd say
19:47:54  <AndreasMadsen>ryah: I'm hoping that a general Q, because I can't answer this satisfactorily.
19:50:21  <ryah>yes, genearl question
19:50:27  <ryah>mmalecki: hm okay
19:50:38  <ryah>mmalecki: will that create a new local branch ?
19:50:51  <ryah>mmalecki: usually i do "curl https://github.com/joyent/node/pull/2388.patch | git am"
19:51:09  <mmalecki>ryah: it'll apply it to your current branch
19:51:16  <mmalecki>yeah, that's pretty much what it does
19:51:18  <ryah>hm
19:52:55  <mmalecki>ryah: you can do it like `hub fetch AndreasMadsen && git checkout cluster2`, I think
19:53:07  <ryah>AndreasMadsen: cluster.worker.send ?
19:53:33  <AndreasMadsen>ryah: If you where just as strict about make jslint as you are about // comment
19:54:15  <AndreasMadsen>ryah: I know but this seams to be the best design ...
19:55:24  <AndreasMadsen>The process.send exist from child_process and we should not overwrite that ...
19:55:49  <ryah>why does cluster.worker.send do that process.send does not?
19:55:55  <ryah>s/why/what
19:56:32  * eddybquit (Read error: Connection reset by peer)
19:56:37  <AndreasMadsen>When the worker do cluster._setupWorker from core/node.js if will create a Worker object assigned to cluster.worker.
19:56:45  <AndreasMadsen>Standby next Q :)
19:57:31  <AndreasMadsen>ryah: We need a way to isolate internal messages and userland messages ...
19:58:20  <AndreasMadsen>The cluster.worker.send create a wrapper object there tell if this is an internal object or not. The process.send just send a json object.
20:04:16  * piscisaureusquit (Ping timeout: 268 seconds)
20:11:18  <ryah>AndreasMadsen: why do we need to isolate internal messages
20:11:40  <ryah>can't we just prefix them with underscore?
20:12:27  <AndreasMadsen>ryah: How prefix?
20:12:46  <AndreasMadsen>Also the wrapper also do callbacks id.
20:12:47  <ryah>{ cmd: '_ident', workerId: 5 }
20:14:33  <AndreasMadsen>ryah: I see but you still have to solve the callback issue. You can't use cluster.worker.send('hallo', callbackFn) since it send a non object.
20:14:56  <ryah>do we need to give people a callback?
20:15:13  <ryah>can't they build that themselves?
20:15:21  <ryah>i want them to be able to use dnode over the IPC
20:16:38  <AndreasMadsen>Also I think its a bad design to have people do if (isObject(msg) && typeof msg.cmd === 'string' && msg.cmd[0] === '_') on every single on('message').
20:17:32  <AndreasMadsen>ryah: I have not thought about dnode
20:17:53  <ryah>what does the callback signify?
20:18:03  <ryah>the remote end sends an ack ?
20:19:16  <AndreasMadsen>Yes
20:19:34  <ryah>is that even useful?
20:19:47  <ryah>tht doesn't mean the remote end has acted on it
20:20:09  <ryah>in fact they might crash directly after sending the ack
20:22:02  <ryah>AndreasMadsen: i think we should match the API for child_process.fork IPC
20:22:03  <AndreasMadsen>ryah: It is very useful internal since you need to know when a suicide state is set, an not just kill it. But the internal callback handling is a little different from the public since it decides when to send callback.
20:22:18  <ryah>we're doing a lot of work to make server.listen() owork transparently
20:22:39  <ryah>i think it makes sense to keep cp.fork and cluster.fork acting similiar
20:22:55  <ryah>i agree that replacing process.send isn't very good...
20:23:45  <ryah>the API surface area is being increased too much here
20:24:06  <AndreasMadsen>ryah: you want userland to just use process.send because cluster.worker.send is a nightmare to type, or what.
20:24:19  <ryah>a nightmare to remember
20:24:47  <AndreasMadsen>I now agree that callback do not need to be supported in userland.
20:28:08  * piscisaureusjoined
20:29:14  <AndreasMadsen>ryah: How are we going to test that process.send will work in every single case. ...
20:29:15  <AndreasMadsen> When i did the cluster.worker.send original it was as you say (but without cmd: '_') ...
20:30:08  <ryah>AndreasMadsen: what do you mean? i think we can just pass a few maessages back and forth to test it
20:30:09  <AndreasMadsen>But there was a lot if issues to fight (example: you could not do .send('string', callback) ).
20:31:05  <AndreasMadsen>At a point I decided to make the codebase simpler by using a wrapper object ...
20:31:34  <ryah>process.send({ message: 1 }) on the server if (message.message == 1) { worker.send({ message: 2}) }
20:31:37  <ryah>etc
20:31:41  <AndreasMadsen>But I do understand you concern.
20:32:35  <ryah>bbl
20:32:35  <AndreasMadsen>ryah: on the server ?
20:32:44  <ryah>AndreasMadsen: s/server/master
20:32:52  <AndreasMadsen>Oh
20:34:31  <AndreasMadsen>As said I agree that we should not support callbacks.
20:36:14  <AndreasMadsen>ryah: the when using a wrapper object we are 100% safe, but when using prefixes I'm worried about what userland modules using prefixes too.
20:36:55  <AndreasMadsen>delete what
20:46:47  * piscisaureusquit (Ping timeout: 252 seconds)
20:47:00  <AndreasMadsen>ryah: I will try modifying the message architecture, we will still keep cluster.worker.send but it will be total equal to process.send so it do not matter. cluster.worker.send will just be isomorfisme.
21:08:31  <ryah>AndreasMadsen: well - start with that
21:08:39  <ryah>AndreasMadsen: we'll argue later about api :)
21:08:48  <AndreasMadsen>Sure :)
21:28:44  * mikealjoined
21:43:38  * AndreasMadsenquit (Remote host closed the connection)
22:24:25  * paddybyersjoined
22:31:42  * luxigojoined
22:37:22  * paddybyersquit (Quit: paddybyers)
22:42:48  * AvianFluquit (Quit: Leaving)
22:45:30  * luxigoquit (Quit: Page closed)
23:59:40  * piscisaureus__joined