[00:05] mikeal has joined the channel
[00:15] mies_ has joined the channel
[00:19] un0mi has joined the channel
[00:22] kriskowal has joined the channel
[00:38] JimBastard: so if the node instance is shutting down assume something like (SIGINT) is it possible to guarantee a POSIX call to complete? 
[00:38] JimBastard: or will the instance shut down before the call is complete?
[00:40] kriskowal: if node behaves as i'd expect (which i don't expect) when you register a SIGINT callback, node would set up a low level SIGINT handler that would arrange for your callback to be called in a future turn.  if that's the case, even if the system interrupt occurs while executing another function, it would be guaranteed to complete.
[00:41] kriskowal: it would be the responsibility of the javascript signal handler to exit gracefully
[00:41] kriskowal: JimBastard^
[00:41] JimBastard: reading
[00:42] JimBastard: so im doing, process.addListener("SIGINT", function(){  application.end();});
[00:42] JimBastard: and in application.end() i have a posix write call that logs the application shutdown
[00:42] JimBastard: in some cases it winds up empty, which it should never be
[00:42] kriskowal: i think you need to explicitly exit in the handler, but that looks right to me.
[00:42] JimBastard: hrmm
[00:43] JimBastard: ill debug more and see if i can catch it then
[00:43] JimBastard: thanks
[00:43] kriskowal: also, i could be wrong about my assumptions about what node does when you add a sigint listener.  i could read the code to verify.
[00:45] kriskowal: looks like node calls down into EV to register the signal handler.  it's very likely that it works as i described.
[00:45] JimBastard: so kriskowal i'm actually already logging all the posix calls, and they arent being called in certain cases
[00:45] JimBastard: it seems that when node is hanging from another error somewhere
[00:46] JimBastard: and i kill it with a CTRL-C, the posix calls dont make it
[00:46] kriskowal: does it actually exit?
[00:46] JimBastard: how do you mean? does node stop?
[00:46] kriskowal: right
[00:46] JimBastard: yeah, ill show you a pastie
[00:46] kriskowal: if you intercept sigint and don't explicitly exit, it should not stop
[00:47] kriskowal: so, if it stops, your handler is probably not getting called
[00:47] isaacs_mobile has joined the channel
[00:47] JimBastard: http://gist.github.com/279107
[00:47] kriskowal: bearing in mind that i'm inferring a lot from experience with signals elsewhere, not from studying node
[00:48] JimBastard: maybe ive fucked up the POSIX calls? kinda a POSIX noob
[00:48] alex-desktop has joined the channel
[00:48] JimBastard: it seems to work in most cases though
[00:49] kriskowal: what are your typical and atypical results
[00:50] kriskowal: if they differ, there's a race condition
[00:50] JimBastard: typical results is the application variable serializing into JSON and storing in a file, atypical is the file getting blanked out
[00:50] kriskowal: and your code sample looks good to me, unless write events call back before the data has been fully flushed and the stream closed
[00:50] JimBastard: it should never be blank, as application always has data in it
[00:50] JimBastard: thats what im not sure about
[00:51] JimBastard: as i am a posix and node noob
[00:51] kriskowal: does node provide an explicit close that returns a promise?
[00:51] JimBastard: no clue
[00:51] JimBastard: i think the error tends to happen when node is already hanging from something else (like not finishing / closing an http request somewhere)
[00:51] kriskowal: probably does.  try putting the exit in a callback for posix.close(fd)addCallback(function () {…
[00:52] kriskowal: 
ACTION forks your gist

[00:52] JimBastard: i'm confused, you mean add a handler to .write for .close?
[00:52] JimBastard: then call exit in .close?
[00:53] JimBastard: ohh thats sweet i didnt know you could fork gists
[00:54] kriskowal: https://gist.github.com/279108/a5c70077e2433b437f8ad40d44452e2ef63e61d8
[00:56] JimBastard: trying it out and getting tacos, brb
[01:00] isaacs has joined the channel
[01:03] mikeal has joined the channel
[01:04] teemow has joined the channel
[01:06] okito has joined the channel
[01:07] JimBastard: so kriskowal i think that might be working....its hard to test as the nature of the issue. i'll let you know if it comes up again
[01:07] JimBastard: thank you
[01:07] kriskowal: np
[01:07] JimBastard: GUESS WHO HAS A NEW GITHUB FRIEND
[01:07] JimBastard: ;-)
[01:17] JoePeck: it looks like the LICENSE was out of date, since parseUri has been gone for a while. It was removed December 18th => http://bit.ly/6e2c4a
[01:17] jamiew_ has joined the channel
[01:17] JoePeck: patch => http://bit.ly/7ARQ8O
[01:30] sprsquish has joined the channel
[01:31] isaacs: JoePeck: thanks!
[01:31] isaacs: i updated the docs and tests, but forgot about hte license.
[01:31] JoePeck: isaacs: sure =)
[01:32] JoePeck: are there easier ways to put up patches? I could have made a github fork, but from following the list I've seen a lot of people using gists
[01:33] r11t has joined the channel
[01:34] isaacs: JoePeck: i'm a fan of having your own fork, and then sending a link to a commit rather than putting the patch on github.
[01:34] r11t has joined the channel
[01:34] isaacs: but ryah's pretty liberal about it, afaict
[01:35] JoePeck: isaacs: okay, I'll do that in the future. I'm still building up some github-fu when working with other people's forks
[01:35] JoePeck: cheers
[01:35] isaacs: suresure
[01:36] r11t has joined the channel
[01:36] isaacs: gist patches are pretty darn easy to apply, though.
[01:36] isaacs: git apply <(curl http://gist.github.com/raw/279121/db762fcdeac4fca58edba99f8a7e5bd4e9104643/gistfile1.diff)
[01:36] JoePeck: heh, I figured people would do that
[01:37] r11t has joined the channel
[01:38] isaacs: JoePeck: http://github.com/isaacs/node/commit/fc30f0f16d10a238f0e07c40e34e10ec0fdb609e
[01:38] r11t has joined the channel
[01:39] JoePeck: isaacs: awesome, thanks =)
[01:39] cloudhead has joined the channel
[01:39] isaacs: sending contribs to the list is pretty good because it at least keeps it all going through one channel.
[01:39] isaacs: (i just sent my commit to the list)
[01:40] inimino has joined the channel
[01:40] JoePeck: k, I figured something this simple could be mentioned in irc, but I see what you mean, with all the forks that are available
[01:41] jamiew_ has joined the channel
[01:46] FoxFurry has joined the channel
[01:47] FoxFurry has joined the channel
[01:48] FoxFurry has joined the channel
[01:48] steadicat has joined the channel
[01:48] isaacs: JoePeck: yeah, if one of us are in here, and you don't feel like doing all the hullabaloo of forking and cloning and pushing etc, then that's fine.
[01:49] FoxFurry has joined the channel
[01:50] FoxFurry has joined the channel
[01:52] FoxFurry has joined the channel
[01:53] FoxFurry has joined the channel
[01:55] FoxFurry has joined the channel
[01:55] FoxFurry has joined the channel
[01:56] FoxFurry has joined the channel
[01:57] FoxFurry has joined the channel
[01:57] teemow has joined the channel
[01:59] teemow has left the channel
[01:59] un0mi has joined the channel
[01:59] jspiros has joined the channel
[01:59] mikeal has joined the channel
[02:02] mikeal has joined the channel
[02:27] mikeal has joined the channel
[02:27] un0mi has joined the channel
[02:39] eviltwin has joined the channel
[02:41] mikeal has joined the channel
[02:48] un0mi has joined the channel
[02:48] inimino has joined the channel
[02:49] inimino has joined the channel
[02:51] inimino has joined the channel
[02:51] cloudhead has joined the channel
[02:52] r11t has joined the channel
[02:52] inimino has joined the channel
[02:52] cloudhead has joined the channel
[02:53] r11t has joined the channel
[02:54] inimino has joined the channel
[02:56] spoob has joined the channel
[03:14] jamiew has joined the channel
[03:27] achew22 has joined the channel
[03:48] steadicat has joined the channel
[03:55] cadorn has joined the channel
[03:55] rauchg has joined the channel
[04:08] JimBastard: ugh not having proper middle ware is kinda annonying right now for this app
[04:08] JimBastard: ive got to keep track of a shit ton of places where the request could end
[04:09] isaacs: joshthecoder: are you joshbuddy?
[04:12] isaacs: guess not.
[04:20] steadicat has joined the channel
[04:22] joshthecoder: isaacs, nope
[04:22] isaacs: kk
[04:23] un0mi has joined the channel
[04:29] rtomayko has joined the channel
[04:47] JoePeck_ has joined the channel
[04:50] JimBastard: inimino you alive
[05:02] sprsquish has joined the channel
[05:12] micheil: JimBastard: node-smtp now sends mail fine.
[05:21] Yuffster has joined the channel
[05:25] JimBastard: nice micheil
[05:25] micheil: yeah
[05:26] JimBastard: whats up with the usage docs
[05:26] JimBastard: ;-)
[05:26] JimBastard: ahh i see now
[05:33] micheil: yeah, no docs.
[05:33] micheil: look at the source, it's all promise based
[05:54] un0mi has joined the channel
[06:31] kriskowal has joined the channel
[06:50] scudco has joined the channel
[06:57] mikeal has joined the channel
[06:57] jed has joined the channel
[07:19] technoweenie has joined the channel
[07:20] jamiew has joined the channel
[07:24] steadicat has joined the channel
[07:51] richtaur has joined the channel
[07:51] richter has joined the channel
[07:54] qFox has joined the channel
[07:56] jamiew_ has joined the channel
[08:19] JimBastard: soo uhh whats up with http://jackjs.org/ and node
[08:25] mikeal has joined the channel
[08:27] jamiew_: hey hey pkg mgmnt
[08:27] jamiew_: re. jack/narwhal.js
[08:28] JimBastard: has anyone tried http://github.com/raycmorgan/gateway yet?
[08:34] mikeal has joined the channel
[08:55] jamiew_ has joined the channel
[09:01] felixge has joined the channel
[09:29] brosner has joined the channel
[09:35] felixge has joined the channel
[09:50] piranha has joined the channel
[10:02] micheil: JimBastard: I think the api would be better as: var builder = new Gateway.Builder();
[10:03] JimBastard: i havent tried it out
[10:03] JimBastard: i dont really have time to mess with middleware right now
[10:03] JimBastard: i think jack just needs to get ported to node
[10:09] JimBastard: micheil did you post the node-smtp update on the mailing list yet?
[10:09] micheil: no, I haven't
[10:10] JimBastard: you should fancy up the readme a little, maybe show some usage
[10:10] micheil: First I really want to get tests working
[10:10] hassox has joined the channel
[10:11] micheil: I just changed the readme to use markdown as a step towards that
[10:11] micheil: github's lagging
[10:11] JimBastard: aye
[10:12] micheil: hmm.. a git client in pure node js?
[10:12] micheil: not possible until ssh is written and until setSecure works on the client
[10:12] micheil: If I knew more about ssl, I'd implement setSecure on the client tcp socket
[10:13] JimBastard: i say push it working with localhost
[10:13] micheil: I'm looking at implementing a schema less datastore soon, which should be fun: http://bret.appspot.com/entry/how-friendfeed-uses-mysql
[10:13] JimBastard: and get people using it
[10:13] JimBastard: the majority of usage will be localhost
[10:13] micheil: true
[10:13] achew22 has joined the channel
[10:13] JimBastard: someone might even come along and help
[10:13] JimBastard: you know
[10:14] JimBastard: ill blog about it and post on hackernews or something
[10:14] micheil: hmm.. I really need to get a jekyll site up..
[10:14] JimBastard: github is fine
[10:14] JimBastard: just get the readme looking fancy
[10:14] micheil: something simple for brandedcode.com instead of that placeholder
[10:14] achew22: Is there a function that will do print_r or var_dump in node.js?
[10:14] micheil: (I'm meaning so that I can blog about it)
[10:15] JimBastard: achew22 where are you debuggint to? the console?
[10:15] DamZ has joined the channel
[10:15] achew22: Yeah, into the console
[10:15] micheil: achew22:sys.puts(sys.inspect(Object))
[10:15] JimBastard: yeah
[10:15] JimBastard: so you could do this
[10:15] JimBastard: sys.puts(JSON.stringify(sys.inspect(Object)));
[10:15] JimBastard: maybe
[10:15] JimBastard: thats a cheat
[10:16] micheil: technoweenie: your semi-colon-less javascript annoys me.. :(
[10:16] JimBastard: you should try out one of the web console debuggers
[10:16] micheil: CI for node :D
[10:16] micheil: that'd be epic.
[10:16] JimBastard: ?
[10:16] JimBastard: achew22 http://maraksquires.com/node_debug/
[10:17] JimBastard: that will give you clickable / async debuggable objects
[10:17] JimBastard: so you can debug.log() large objects and click through to inspect them
[10:17] JimBastard: in a browser
[10:17] JimBastard: sys.puts will only take you so far
[10:19] micheil: what are simplex and picard? could someone fill in there descriptions? http://wiki.github.com/ry/node/modules
[10:19] JimBastard: ummm
[10:20] JimBastard: i dunno
[10:20] JimBastard: clever framework gets clever framework name + 1
[10:22] achew22: micheil: They exist in the section labeled "Web frameworks"
[10:23] micheil: achew22: so they should be web frameworks?
[10:23] micheil: 
ACTION organised that page, but wants every module to have a short description with it

[10:25] micheil: *sigh* Hashlib looks great, although, I'd rather use a js module I can submodule and require, without compiling.. 
[10:26] JimBastard: yeah micheil that is a whole other can of worms
[10:27] JimBastard: i was thinking about doing continious intergration instead of hot swapping of modules and shit
[10:27] micheil: that's my biggest issue with the difference between C++ / JS modules
[10:27] JimBastard: when you start doing dynamic modules / requiring you run into all the fun problems with dynamic langauges
[10:27] JimBastard: depenency hell
[10:28] JimBastard: dependency hell
[10:28] micheil: I mean, I could probably definitely get some large speed improvements in node-smtp by moving Queue.js to be a c++ module, but I don't want deps.
[10:28] JimBastard: im sure someone wrote that in c++ already?
[10:28] micheil: I guess I'll wait until we have a solid way to consistently download / build / package things
[10:28] micheil: nup
[10:28] micheil: Queue is an array based runner
[10:29] JimBastard: no idea
[10:29] JimBastard: are you using that in node-smtp?
[10:29] micheil: so.. I push data to the queue, then it immediately triggers a process() 
[10:29] micheil: once the process feels it has a packet, it emit's an event
[10:30] micheil: yeah, Queue.js is at the core of node-smtp
[10:30] micheil: brb, getting my glasses.
[10:30] micheil: back
[10:31] micheil: I use queue as a way to ensure that whole packets / commands are received
[10:31] micheil: I may change it to be string based, rather then array based though
[10:33] JimBastard: have you tried running any blast emails?
[10:33] micheil: as in?
[10:33] JimBastard: like 1000 emails at once?
[10:33] JimBastard: 10,000? etc
[10:34] micheil: you can't
[10:35] JimBastard: so whats the limit
[10:35] micheil: because the smtp protocol is send/receive based
[10:35] JimBastard: etc
[10:35] JimBastard: durrrr
[10:35] JimBastard: work with me here
[10:35] JimBastard: a mail queue
[10:35] JimBastard: of 10,000
[10:35] JimBastard: to be blasted
[10:35] micheil: how ever, you could more then likely make a class to instantiate many clients
[10:35] micheil: I might make that into mailer.js
[10:35] JimBastard: well assuming you send them one at a time single threaded
[10:36] micheil: yeah
[10:36] JimBastard: will it work?
[10:36] micheil: the way it'd work is because all of node-smtp is using promises
[10:36] JimBastard: i would add a test case for like 1,000 
[10:36] micheil: promises and event emitters
[10:36] JimBastard: just to make sure smtp.client doesnt shit itself
[10:36] micheil: it shouldn't
[10:36] JimBastard: word
[10:37] micheil: due to the way the SMTP protocol is, it'd have to have either "threads" or a queue of impatient hamsters
[10:37] micheil: by "threads" I mean, it'd be multiple clients
[10:38] JimBastard: "workers"
[10:38] micheil: yeah
[10:38] JimBastard: thats the word
[10:38] micheil: same idea
[10:39] JimBastard: doesnt have to be that fancy i think
[10:39] micheil: so, lib/mailer.js would include a method for sending bulk / a queue
[10:39] JimBastard: it should probaly queue by default
[10:39] JimBastard: and everything goes through queue
[10:39] micheil: you could set into it the number of workers it's allowed to spawn, and then etc.
[10:39] JimBastard: yeah
[10:39] JimBastard: im sending your emails in javascripts
[10:39] JimBastard: awesome
[10:40] JimBastard: fucking awesome
[10:40] micheil: if you set it to spawn only one worker, then it'll queue the messages
[10:40] micheil: you could also make use of Queue.js in there
[10:40] JimBastard: thats what im saying
[10:40] micheil: Queue.js is much like a data pump
[10:40] JimBastard: since its just a process right
[10:40] JimBastard: or a promise
[10:40] JimBastard: just register it
[10:41] micheil: it's not a process or promise, it's a worker
[10:41] JimBastard: aight 
[10:41] JimBastard: not really sure what queue.js does
[10:41] micheil: have a read: http://github.com/miksago/Queue.js
[10:41] JimBastard: just read it
[10:42] micheil: right, so if you look at the example implementation there, I can replace the worker to do the sending of emails.
[10:42] JimBastard: so
[10:42] JimBastard: the queue (right now) will fire everything you register in it sequentially?
[10:42] micheil: well, you could do: new SMTP.workerClient(port, host, threads);
[10:42] micheil: yes
[10:43] micheil: how ever, that acts as a queue
[10:43] JimBastard: so when you email.send() it will add a job to the queue?
[10:43] micheil: we spawn many instances of SMTP.client() based on the threads
[10:43] micheil: yeah
[10:43] JimBastard: that sounds pimp
[10:43] JimBastard: if it works how i think it does
[10:43] JimBastard: im a little retarded
[10:43] micheil: it'll add it to the global queue, and then the first free SMTP.client will send the message
[10:48] micheil: woot!
[10:48] micheil: got subjects working..
[10:48] micheil: at some stage I broke them, but I got them working again
[10:49] JimBastard: im probaly gonna install your build in the next few days
[10:49] JimBastard: i have to do a friend import feature which will require sending out like 500 emails
[10:49] JimBastard: i dont think my code will handle it
[10:49] micheil: heh heh
[10:49] micheil: just use it as a git submodule
[10:49] JimBastard: sup
[10:50] un0mi has joined the channel
[10:50] micheil: "sup" ?
[10:51] micheil: JimBastard: as a new form of ypu?
[10:51] micheil: *yup
[10:51] JimBastard: as in
[10:52] micheil: btw, what are you building with node JimBastard ?
[10:52] JimBastard: Kind sir, I am intrigued by what you are talking about and wish to inquire further about what a git submodule is and how it would be applicable to this situation in hence we are currently talking about.
[10:52] JimBastard: or
[10:52] JimBastard: sup
[10:53] JimBastard: custom application
[10:53] micheil: okay, are you managing with git?
[10:53] JimBastard: that requires large amounts of http requests and open requests
[10:53] JimBastard: yeah kinda
[10:53] JimBastard: im pretty bad though
[10:53] micheil: if you are: git submodule add git://github.com/miksago/node-smtp.git
[10:53] JimBastard: what will that do
[10:53] micheil: in the root of your application
[10:53] micheil: actually.
[10:54] micheil: git submodule add git://github.com/miksago/node-smtp.git vendor/node-smtp
[10:54] micheil: will add a node-smtp directory in the vendor directory of your root
[10:55] micheil: then when you do: git submodule update
[10:55] micheil: you'll automatically get the latest version of node-smtp
[11:01] JimBastard: word
[11:01] JimBastard: ill check it out later for sure
[11:02] JimBastard: i gotta build the logic to import before i can send out 500
[11:02] JimBastard: :-)
[11:02] micheil: heh heh
[11:05] micheil: JimBastard: It'll be a while before the API to Mailer is frozen, so watch out with thT
[11:05] micheil: *that
[11:06] JimBastard: well you know
[11:06] JimBastard: iterate often and shit
[11:06] JimBastard: right now i can send some emails
[11:06] JimBastard: way better then none
[11:07] achew22: Are there enough people working on node based projects to identify trends like pack leaders for most accepted webframework or anything like that?
[11:10] JimBastard: github followers
[11:10] JimBastard: express
[11:11] JimBastard: for most of what im doing, modules are king
[11:11] JimBastard: no point having a glue if you have nothing to glue together
[11:12] JimBastard: right now im working with session.js, application.js, route.js, email.js , debug.js , mime.js, mustache.js 
[11:12] JimBastard: http://github.com/visionmedia/express
[11:13] achew22: is it preferred to make a system wide folder that contains your libraries or is it better to git submodule them?
[11:14] JimBastard: i dunno
[11:20] JimBastard: mmmm break time
[11:20] JimBastard: bbl
[11:26] micheil: achew22: depends, I stick to putting any third party code into /lib/vendor
[11:26] micheil: to say, I may or may not have written this, but I use it
[11:26] micheil: and then where a git repo is available, I use submodules
[11:49] BBB has joined the channel
[11:56] un0mi has joined the channel
[12:30] tisba has joined the channel
[12:43] DamZ has joined the channel
[13:01] felixge has joined the channel
[13:12] felixge has joined the channel
[13:19] markwubben has joined the channel
[13:21] lifo has joined the channel
[13:24] rolfb has joined the channel
[13:38] micheil: man, that is so the next protocol I'll be implementing. MPD
[13:39] mediacoder: you should try SPDY :-)
[13:43] micheil: I just "hacked" my brother's arch box because he had MPD running on localhost instead of loopback
[13:43] micheil: mediacoder: SPDY looks like something google related, right SPDY?
[13:44] micheil: faster http protocol or something
[13:44] mediacoder: yea, not at all mail related..just saw your love towards protokol implementations and rfcs :-)
[13:44] micheil: (besides, MDP uses a very similar reply structure to telnet)
[13:44] richter has joined the channel
[13:44] micheil: not tlenet.
[13:44] micheil: SMTP
[13:46] micheil: but yes, protocol documentation / RFC's do make implementation much easier
[13:49] micheil: mediacoder: I'm not sure, but looking very quickly, I probably can't implement SPDY, as it requires SSL
[13:51] micheil: mediacoder: how much do you know about this SPDY protocol?
[14:03] micheil: hmm.. I reckon I can get a baseline implementation of MPD in about 2 days.
[14:31] mediacoder: micheil: nothing :-) i was rather kidding, but anyways SPDY might be interesting for node..being even more pioneering
[14:32] micheil: mediacoder: If you've seen my latest tweet, then you'll know I'm up for a challenge on protocols.
[14:33] micheil: http://github.com/Miksago/node-mpd-client/
[14:33] micheil: it has no code yet, but shouldn't take me too long to spec out
[14:33] micheil: although, I'll need to write a mock server for it first.
[14:34] micheil: it's not as documented as SMTP, although, it does have some documentation
[14:35] micheil: either that or fix up my pc with arch
[14:42] micheil: woot. finally found some screws for my old HDD..
[14:42] JimBastard has joined the channel
[15:05] micheil: hmm..
[15:07] alex-desktop has joined the channel
[16:03] felixge has joined the channel
[16:20] felixge: has anybody used posix.sendfile to stream large files with node yet?
[16:22] inkubus08 has joined the channel
[16:31] felixge has joined the channel
[16:33] sprsquish has joined the channel
[16:35] technoweenie: micheil: your semi-colon-full javascript annoys others too :)
[16:35] micheil: technoweenie: :P
[16:35] technoweenie: i personally don't care, i just figure there's no chance my js will be run in a browser so i dont mind if i leave semicolons out
[16:35] micheil: technoweenie: the only reason it annoys me is because one misplaced something and the js won't run at all
[16:36] micheil: odd.. hmm..
[16:36] micheil: I got a hang up on something that shouldn't have hung up
[16:37] micheil: technoweenie: any experience with MPD by any chance?
[16:38] technoweenie: no... unless you mean uh multiple personality disorder?
[16:38] felixge: looking for a metaphor for serving static files, anybody?
[16:40] micheil: Music Player Daemon
[16:41] bryanl has joined the channel
[16:41] micheil: felixge: this is worse then a crossword
[16:41] felixge: micheil: why? Because I'm not limiting you to a certain amount of letters? :)
[16:42] micheil: yeah
[16:42] technoweenie: well maybe you should, it'd be another clue
[16:42] micheil: ;P
[16:42] felixge: I want to do a node.js module for serving files but 'static' is a reserved keyword in JS ... 
[16:42] steadicat has joined the channel
[16:42] micheil: statik
[16:43] micheil: ?
[16:43] technoweenie: wow, it is?
[16:43] micheil: possiblylargeandnotmovingfile
[16:43] micheil: j/k
[16:43] micheil: yeah, Java's reserved keywords are reserved in javscript
[16:43] micheil: *javascript
[16:44] felixge: Not sure if v8 would complain, but either way it wouldn't be a very safe choice
[16:44] felixge: I like 'statik' so :)
[16:44] felixge: I'll use that
[16:46] micheil: hmm..
[16:46] micheil: 
ACTION is hooking up a library to allow node to control a music player :D

[16:46] micheil: I'm also experimenting with various closed / eof errors
[16:47] Booster has joined the channel
[16:47] JoePeck has joined the channel
[16:48] felixge: micheil: sounds like fun :)
[16:48] okito has joined the channel
[16:48] micheil: felixge: it is, although, I don't have the music player available to me for me to test
[16:48] micheil: so I'm writing a "mock server"
[16:49] felixge: micheil: now why would you work on that?
[16:50] micheil: because, if I get one mock server working, I should easily be able to write on for node-smtp
[16:50] micheil: ;P
[17:04] jan____: felixge: what do you need the metaphor for?
[17:04] jan____: project name?
[17:05] felixge: jan____: yeah
[17:05] felixge: I want to write a module that detects mime type based on the file extension and properly streams the file to the client
[17:05] felixge: so you can easily build a node-app that also serves its own static assets from a webroot
[17:06] micheil: nice one felixge 
[17:06] micheil: felixge: make sure you add it to the wiki page of modules
[17:07] felixge: micheil: will do
[17:13] jan____: felixge: Paperboy
[17:14] r11t has joined the channel
[17:14] felixge: jan____: I guess that could work :)
[17:14] steadicat has joined the channel
[17:14] jan____: :)
[17:15] jan____: 
ACTION prepares an Awesome_Paperboy

[17:16] felixge: still kinda hard to make the metaphor go full circle
[17:16] felixge: but I think it will do
[17:16] jan____: :)
[17:17] tisba has joined the channel
[17:17] felixge: paperboy.deliver(filename, res).before(fn).after(fn).otherwise(fn);
[17:17] felixge: I think will be the API
[17:17] jan____: :)
[17:18] jan____: add .whatever(fn) for good measre
[17:18] jan____: hmm
[17:18] jan____: try {} catch() {} whatever {}
[17:18] felixge: jan____: when would whatever fire?
[17:18] jan____: whenever ;D
[17:19] jan____: it makes more sense in a try-catch as a replacement for finally
[17:20] Booster has joined the channel
[17:21] stevestmartin has joined the channel
[17:27] rolfb has joined the channel
[17:36] micheil: felixge: nice api you've got there
[17:36] micheil: it'd be interesting if I could do: paperboy.deliver("./public/*", res);
[17:37] micheil: although, that's probably something I should handle
[17:37] micheil: horaay! just got a pump based mock server and client working for node-mpd-client
[17:42] bryanl has joined the channel
[17:42] micheil: hmm.. this mockServer + client works pretty good
[17:42] micheil: I've got it such that if either hangs up, the other knows
[17:42] felixge: hm I think the 'drain' event is broken
[17:43] micheil: if the client hangs up, the server stays online for 10 seconds and then exits
[17:43] micheil: if the server hangs up, then the client quits cleanly
[17:43] micheil: felixge: I think there was a bug about that.
[17:44] felixge: micheil: was it fixed? or delayed for the net2 branch merge?
[17:44] micheil: I'm not sure
[17:44] micheil: ryah: would have to answer that..
[17:44] micheil: I guess.
[17:46] micheil: I should head to bed.. it's 4:40am
[17:46] BBB has joined the channel
[17:47] micheil: kennethkalmer: I'll probably rewrite some of the core socket parts of node-smtp later on today, based on what I've just learnt from these tests work mockServer / client
[17:47] micheil: G'night folks. felixge could you dm me on github when you release paperboy?
[17:48] micheil: (Miksago)
[17:48] felixge: micheil: k, will try to remember
[17:48] micheil: cheers
[17:48] micheil: I'll probably find it on github tomorrow or something
[17:48] micheil: when I read back to what I was doing now.
[17:49] micheil: night.
[17:51] kennethkalmer: micheil: looking forward to it, good night
[17:52] micheil: kennethkalmer: yeah, it's basically going to be a better queue/pump working, and have a better hangup / close connection system
[17:52] kennethkalmer: now go get some rest :)
[17:53] micheil: I'm still not sure if I should go to using string based pump over array based.. gotta sort that out and figure out which is quicker.
[17:53] micheil: night.
[17:58] rictic has joined the channel
[18:15] pdelgallego has joined the channel
[18:18] bry has joined the channel
[18:27] cloudhead has joined the channel
[18:38] rictic has joined the channel
[18:49] technoweenie: hey when does a node.js call process.exit()
[18:49] technoweenie: is there a way to see whats running so i can see whats hanging the process
[18:55] felixge has joined the channel
[19:09] sztanpet has joined the channel
[19:22] jamiew has joined the channel
[19:25] jed__ has joined the channel
[19:30] erichocean has joined the channel
[19:34] felixge: Released paperboy, a little #nodejs module for delivering static files: http://bit.ly/6DGVhF
[19:34] jan____: woot
[19:35] jan____: felixge: copy and paste fail in the License section of the readme
[19:35] felixge: I wouldn't use it for huge files so. Since I can't seem to get a 'drain' event for the tcp connection there is a good chance huge files might hock memory in the buffers
[19:35] jan____: also, no naming credit for me :D
[19:35] felixge: oh
[19:35] felixge: shit
[19:35] felixge: :)
[19:35] felixge: now you'll get two credits
[19:35] jan____: hehe
[19:37] felixge: jan____: both fixed
[19:37] felixge: :)
[19:37] felixge: sorry
[19:38] mediacoder: felixge: why didnt you go the sendfile way? :-) 8i was curious about its usage so i was looking forward to paperboy :-))
[19:38] felixge: mediacoder: because sendfile seems broken. It only streams parts of the file for me.
[19:39] mediacoder: ah, i see
[19:39] technoweenie: hey i have a question about conventions
[19:39] technoweenie: in a module i want to expose a class
[19:39] felixge: mediacoder: and since I don't have a drain event I cannot find out when the headers have been send
[19:39] technoweenie: exports.MyClass = MyClass = function(...) { 
[19:39] technoweenie: or
[19:40] technoweenie: exports.create = function() { return new MyClass(arguments) }
[19:40] technoweenie: var MyClass = require('my-class') ; var obj = MyClass.create(...)
[19:40] felixge: technoweenie: I recommend: var MyClass = exports.MyClass = function(...) {
[19:41] felixge: I think this works just as well so:
[19:41] felixge: exports.MyClass = function MyClass(...) {
[19:41] felixge: maybe I should use that
[19:41] technoweenie: yea i like the idea of just exposing the class instead of adding some little api on the exports object (the exports.create thing)
[19:41] technoweenie: felixge: nope...
[19:41] technoweenie: "exports.TwitterNode = function TwitterNode(options) {" gives me ReferenceError: TwitterNode is not defined
[19:42] technoweenie: because right below that function i'm doing: TwitterNode.prototype.track
[19:42] technoweenie: TwitterNode.prototype.track = function(...) {...}
[19:42] jan____: yeah, TwitterNode is not getting registered with the global object
[19:43] technoweenie: yea thats what i thought.  im going with felixge's first suggestion
[19:43] technoweenie: but my prototype declarations dont seem to be sticking
[19:43] technoweenie: hrm
[19:45] technoweenie: AH sys.inherits(TwitterNode, process.EventEmitter)
[19:46] technoweenie: that clears all my prototype methods
[19:46] inimino: felixge: I have sendfile working reliably, but there's some ugliness involved
[19:47] mediacoder: felixge likes it dirty :-) (id be interested, aswell, if you have it to show :-)
[19:47] jan____: lol
[19:48] inimino: haha
[19:49] mediacoder: (sorry for that one, felixge ;-))
[19:49] inimino: mediacoder, felixge: http://www.pastie.org/782207
[19:49] felixge: technoweenie: yeah, you have to call sys.inherits *before* adding your stuff to the prototype
[19:49] technoweenie: yea i see that now :)   http://github.com/technoweenie/twitter-node/commit/b2ccdb0008dcfaca7cf448fcab8cf608a06138a1
[19:50] inimino: I haven't benchmarked it against the regular way
[19:50] felixge: inimino: are you using that for http?
[19:50] felixge: inimino: how do you send the headers out and make sure they got send before invoking sendfile?
[19:50] jan____: technoweenie: oh hey, do you work with mattly?
[19:50] technoweenie: yea
[19:51] jan____: cools!
[19:51] inimino: felixge: yes
[19:51] technoweenie: yea i didnt realize you were the couchdb jan until #redis a few minutes ago :)
[19:51] jan____: :)
[19:51] jan____: seems like the cool kids all hang out in the same channels :)
[19:52] technoweenie: heh
[19:52] inimino: felixge: oh, there's some ugliness there too
[19:52] inimino: 
ACTION prepares a second paste...

[19:52] inimino: http://www.pastie.org/782212
[19:52] felixge: who needs redis if they can get dirty? :)
[19:53] felixge: inimino: you got a race condition there
[19:54] felixge: inimino: your code only works because posix.open() is usually slower then your connection.send() for the headers
[19:54] felixge: I think
[19:54] inimino: hm, you sure?
[19:54] felixge: unless connection.send() does some buffering of its own thats honored by sendfile
[19:55] felixge: inimino: well, try do the send() after posix.open() finishes and invoke the sendfile right after that
[19:55] inimino: I thought connection.send just synchronously dumps it into the kernel's send buffer for the socket
[19:56] felixge: inimino: looking at the code it seems you are right
[19:56] felixge: hm
[19:58] inimino: the ugliest thing about this is the if(e.message=='Resource temporarily unavailable'){...
[19:58] inimino: which you do need, without that you'll only get partial sends
[20:00] felixge: inimino: I guess I'll wait for the net2 branch before switching to sendfile
[20:00] felixge: I already have one dirty project : )
[20:01] mediacoder: hehe
[20:02] inimino: hehe, yeah, I'm ditching this too in favor of just using plain vanilla streaming
[20:03] isaacs has joined the channel
[20:03] felixge: inimino: yeah, once 'drain' is working again it should be efficient "enough" for small web projects
[20:04] felixge: I'd go with nginx/lighttpd for more serious file serving needs anyway
[20:04] felixge: but I would like to do some benchmarking against them
[20:04] inimino: well, I may benchmark the two eventually
[20:04] felixge: yeah
[20:04] inimino: I'd kind of like to get sendfile working properly
[20:04] felixge: that'd be fun
[20:04] felixge: If sendfile works I think we would have a pretty good chance of being at least as fast as nginx
[20:05] felixge: I would imagine nginx to do much more when serving a regular file than node will on the minimum setup
[20:05] inimino: but it's not earning a spot on my TODO list just yet
[20:05] inimino: yeah, I'd say so
[20:05] inimino: yes, especially with fast routing :-)
[20:08] inimino: basically you have an incoming HTTP request going through http_parser, some request routing/dispatch, and then some JavaScript that's already been compiled to machine code by V8, kicking off some sendfile() and the kernel does the rest
[20:08] inimino: should be about as fast as anything else out there
[20:10] felixge: inimino: yeah, if that's the case node could actually replace traditional web servers all together for most parts :)
[20:10] felixge: we just need an async fcgi module
[20:13] un0mi has joined the channel
[20:13] inimino: felixge: yep
[20:13] felixge: inimino: ryan is going to add support for passing sockets via sockets in net2
[20:14] felixge: this means you can have multiple web workers handling connections
[20:14] felixge: without occurring any communication overhead 
[20:14] inimino: yeah
[20:14] felixge: so sweet
[20:14] inimino: yeah, that will be awesome
[20:14] felixge: :)
[20:14] felixge: Node.js based load balancers would be awesome as well
[20:15] felixge: as you could trivially hack in any logic you need for sessions and stuff
[20:15] inimino: yep
[20:16] Booster has joined the channel
[20:19] jed__ has joined the channel
[20:21] DamZ has joined the channel
[20:23] felixge: isaacs: is npm working right now?
[20:23] isaacs: felixge: define "working"?
[20:23] isaacs: i'm kinda blocked atm
[20:24] isaacs: working on getting unblocke.d
[20:24] felixge: isaacs: compatible with the latest version of node.js 
[20:24] felixge: isaacs: just saw your pull request
[20:24] isaacs: felixge: my npm master is compatible with latest node.
[20:24] isaacs: and it's going to be using commonjs-compatible package.json, extended with the "main" field.
[20:25] isaacs: main = "package entry point module", or "what you get when you do require(package)"
[20:25] felixge: cool
[20:25] felixge: :)
[20:25] isaacs: it can right now install a package if you give it a tarball and the name.
[20:25] isaacs: it doesn't do versioning yet, or install-by-name-only
[20:25] isaacs: since there's no registry, and i am 100
[20:25] isaacs: 5
[20:25] felixge: I really really want a package mamanger
[20:25] felixge: :)
[20:25] isaacs: since there's no registry, and i am 100% not going to have the package catalog live inside npm
[20:26] isaacs: it'll be a bit slower install-time to use a hosted package registry rather than a local one, but we can add caching, and it's worth it for the distributed factor.
[20:26] felixge: yeah
[20:28] isaacs: how's node-dirty coming?
[20:28] bryanl has joined the channel
[20:29] isaacs: i'm planning on using that or some other kind of uber-simple key-value store.
[20:29] isaacs: dirty looks like it hits exactly the features i need for the npm registry, but without any of the features i don't need.
[20:31] jan____: felixge: looks like I can use dirty to implement awesome's store.js.
[20:31] jan____: on the other hand...I already have store.js :)
[20:32] felixge: jan____: its up to you if you want to be dirty :)
[20:34] jan____: felixge: hahaha :D
[20:34] jan____: felixge: I might want to argue that store.js is very dirty indeed
[20:34] erichocean has joined the channel
[20:34] felixge: jan____: actually it would depend on how much of dirty you need
[20:34] felixge: need to look at store.js
[20:35] jan____: felixge: so far it is just an array of objects
[20:35] felixge: jan____: any disk persistence?
[20:35] jan____: and tons of helper methods for commands to operate on these.
[20:35] jan____: felixge: yeah, flush to JSON file
[20:35] felixge: k
[20:35] jan____: much like redis
[20:36] felixge: well I would argue that dirty is not meant to write something like redis
[20:36] felixge: for two reasons
[20:36] felixge: a) Redis already exists
[20:36] felixge: b) I believe it is stupid to send trivial questions like "what is the value of key x" over a network : )
[20:37] jan____: the first 50 LoC of store.js are all I need from Dirty
[20:37] jan____: a) it is extra infrastructure
[20:37] felixge: I want dirty to be the base for building network databases that answer real world questions directly via  a REST API
[20:38] jan____: b) remote, in-memory data structures are sexy
[20:38] felixge: jan____: I'd say keep what you have
[20:38] jan____: yea
[20:38] felixge: I'd be really interested in seeing a benchmark for awesome :)
[20:38] felixge: I got dirty up to 700k writes / sec
[20:38] felixge: but its cheating without networking
[20:39] jan____: I ran some of the redis benchmarks 
[20:39] jan____: as much as were supported by the subset of commands I support
[20:39] jan____: awesome was ca 2x slower than redis
[20:39] jan____: I hoped for 10x
[20:40] felixge: wow
[20:40] felixge: that would be fantastic
[20:40] felixge: I'm wondering how much net2 will make a difference
[20:40] jan____: yeah, busy on getting the rest fixed up to make real tests
[20:40] felixge: yeah
[20:41] felixge: you probably need to do some profiling anyway, its very easy to loose performance in high req/sec scenarios due to stupid things
[20:42] felixge: at least I didn't realize how I lost 600k writes / sec for quite a while *g*
[20:43] felixge: 
ACTION wishes there was a profiler for node/v8 yet

[20:43] jan____: lol
[20:43] jan____: yeah, I have tons of inefficiencies in there
[20:44] felixge: (hint: I was flushing more often than I thought I was which killed performance)
[20:44] jan____: most importantly, the operations don't have redis' big-o guarantees yet, since I haven't analysed the underlying implementation characteristics.
[20:45] felixge: jan____: does that matter for < 10 GB Ram?
[20:45] felixge: :)
[20:45] jan____: so there's ton of O(N) or worse in there where things could be O(log n) or O1)
[20:45] jan____: felixge: the more CPU I burn on that the less reqs I can handle :)
[20:45] jan____: (so no) :)
[20:45] felixge: I have to do some analysis for that for dirty as well
[20:45] felixge: I'd like to say dirty has O(YEAH) characteristics 
[20:46] jan____: HAHAHA
[20:46] jan____: awesome
[20:46] jan____: in the pretext, nebulously define E=... A=... Y=... and H=...
[20:46] jan____: and conclude the performance is O(YEAH)
[20:47] jan____: I'll do that :D
[20:47] jan____: felixge: which brings up the topic of missing implementation details in the API docs of all the JS engines
[20:47] jan____: granted it is unusual, but redis has it :)
[20:47] felixge: jan____: yeah I think you'll have to reverse engineer these
[20:48] felixge: john published this profiler thingy that kinda does that somewhat I think
[20:48] felixge: not sure if it was good enough so
[20:49] jan____: felixge: I'm hoping somebody else could do the gneering
[20:49] felixge: hehe
[20:49] jan____: open souce!
[20:49] felixge: ok, gotta sign off now - pretending to participate in a real world game of cards
[20:50] jan____: gnight!
[20:50] felixge: jan____: anything on the radar next week that could make for an excuse for hanging out with you guys
[20:50] jan____: (what game?)
[20:50] felixge: ?
[20:50] felixge: it was really fun at the js meetup
[20:50] felixge: jan____: romee
[20:50] jan____: felixge: The Gaggia is back in the office, so we have awesome coffee, you're welcome any time
[20:50] felixge: *spellfail
[20:50] jan____: darn, looking for Doppelkopf folks
[20:51] felixge: I'll play anything really
[20:51] felixge: : )
[20:51] felixge: Luck beats skill anytime
[20:51] felixge: :)
[20:51] felixge: 
ACTION really leaving now, gn8 folks

[21:15] scudco has joined the channel
[21:38] jamiew has joined the channel
[21:45] stevestmartin has left the channel
[22:00] sudoer has joined the channel
[22:00] Booster has joined the channel
[22:11] un0mi has joined the channel
[22:12] bryanl has joined the channel
[22:13] aho has joined the channel
[22:19] Booster has joined the channel
[22:20] Yuffster has joined the channel
[22:22] inkubus08 has joined the channel
[22:22] tlrobinson has joined the channel
[22:22] inkubus08 has joined the channel
[22:22] sprsquish has joined the channel
[22:47] r11t has joined the channel
[22:50] jan____: the forced order of declarations is pretty annoying, what am I doing wrong?
[22:51] inimino: declarations?
[22:51] r11t_ has joined the channel
[22:52] jan____: function foo() { bar(); } function bar() { ... }
[22:52] jan____: won't work
[22:52] jan____: I need to define bar first
[22:53] aho: and that's a big problem?
[22:55] inimino: jan____: nonsense, it works perfectly
[22:55] inimino: jan____: let me guess, you're using JSLint?
[22:56] gbot2 has joined the channel
[22:56] inimino: if so, that's definitely what you're doing wrong
[22:57] ibolmo has joined the channel
[22:58] ibolmo: can someone tell me if node.js' assert.ok returns anything?
[22:58] aho: i for one really like jslint
[22:58] ibolmo: rather.. if it returns assert again?
[22:58] gbot2 has joined the channel
[22:58] ibolmo: in other words: can I do: assert.ok(true).ok(true)...
[22:59] isaacs: jan____: you can call functions before they're defined if you're defining them as named function declarations rather than var foo = function
[22:59] isaacs: an odd quirk of the language leading to some strange edge cases and utterly confounding ie<=7
[22:59] gbot2 has joined the channel
[23:02] Booster has joined the channel
[23:03] mikeal has joined the channel
[23:06] jan____: inimino: not jslint.
[23:06] jan____: isaacs: that might be it. I have this.method = function() declarations
[23:06] jan____: sorry I wasn't exact
[23:07] isaacs: yep, that'd do it
[23:07] hassox has joined the channel
[23:07] jan____: fucksticks
[23:07] jan____: it's annoying
[23:07] isaacs: i generally like naming functions, and caling them by name, just to avoid that situation.
[23:07] inimino: jan____: ah, what's giving the issue?
[23:07] inimino: jan____: ah...
[23:07] jan____: well, I'm building an object ;)
[23:07] inimino: yes
[23:07] isaacs: this.bar = bar;  function bar () { .. }
[23:07] ibolmo: I'm emulating the assert module. Does 'ok' return the instance, or assert? e.g. assert.ok(true).ok(true)...
[23:07] jan____: isaacs: I'll five that a shot
[23:07] hassox: morning ppls
[23:07] bryanl has joined the channel
[23:08] isaacs: jan____: but you'll still have to not call this.bar until it's assigned.
[23:08] isaacs: you could always do bar.call(this, 1,2,3) or bar.apply(this, [1,2,3])
[23:08] isaacs: hassox: mornin, mate
[23:08] inimino: jan____: no reason you can't use function declarations and then build the object at the end
[23:08] hassox: hey isaacs  :)
[23:09] jan____: inimino: no reasons beside silly bookkeeping
[23:09] jan____: anyway, off to workarounds
[23:09] inimino: jan____: or return {foo:foo,bar:bar,baz:baz}; function foo(){...} function bar(){...} function baz(){...}
[23:09] jan____: thanks for all the suggestions :)
[23:10] isaacs: jan____: be wary of saying "grr, you can't <blah blah blah> in javascript!" around here. you'll be flooded with suggestions ;)
[23:11] jan____: isaacs: I wouldn't ask here if that wasn't a desired outcome :)
[23:11] inimino: hehe
[23:11] inimino: grr, you can't calculate the busy beaver in JavaScript!
[23:11] jan____: you could... wait a minute!
[23:11] jan____: nearly got me there
[23:12] inimino: hehe
[23:12] ibolmo has left the channel
[23:28] jan____: hehe now this: a local variable defined before the function that uses it, but after the function that calls the function that uses it, causes that var to be undef
[23:46] inimino: jan____: if it's called before the definition is reached, yes
[23:46] jan____: yeah
[23:46] jan____: sucks :)
[23:49] erichocean has joined the channel
[23:51] technoweenie: damn, async stuff is so hard to test right
[23:51] technoweenie: hassox: hey did you see my twitter-node project
[23:51] hassox: technoweenie: yes
[23:51] hassox: well
[23:52] hassox: I saw the repo but I didn't get a chance to go through it
[23:52] technoweenie: i just made it a single event emitter that you can tie into whatever
[23:53] hassox: :)
[23:53] hassox: awesome
[23:53] hassox: the api looked good :)
[23:53] hassox: it's pretty awesome that it can be done :)
[23:54] hassox: did you see I updated chain?
[23:55] technoweenie: no, i couldnt figure out what it was for
[23:55] technoweenie: i'm sticking to the bare metal on this crap before jumping into all these extractions people are writing
[23:56] technoweenie: twitter-node has no dependencies, and only uses a base64 module i ripped off some dudes presentatin
[23:56] hassox: technoweenie: :(
[23:56] hassox: dox are my worst thing
[23:56] hassox: it's a very simple rack level thing to build components on
[23:56] technoweenie: i'm getting a good feel for js again, so i'll probably take another look after i've got waldo in production
[23:57] hassox: it uses async methods to move from one middleware to another
[23:57] technoweenie: ah
[23:57] technoweenie: yea i used a 'simple rack level thing' for a ruby http connection lib
[23:57] technoweenie: its a great pattern
[23:57] technoweenie: i just cant visualize in my head how that might work w/ async stuff yet
[23:58] hassox: http://github.com/hassox/chain/blob/master/API.textile
[23:58] hassox: there's lots of options for _how_ to implement it
[23:59] hassox: but basically, you don't want to use callstacks to move about
[23:59] technoweenie: oh is chain for http servers