00:30:26  * `3rdEdenquit (Quit: Zzzzzz)
00:36:24  * mikealjoined
00:54:03  * nrajlichjoined
00:58:43  * chilts_joined
00:59:00  * ryah_joined
01:03:52  * TooTallNatequit (*.net *.split)
01:03:53  * chiltsquit (*.net *.split)
01:03:53  * ryahquit (*.net *.split)
01:19:20  * nrajlichchanged nick to TooTallNAte
01:19:24  * TooTallNAtechanged nick to TooTallNate
01:29:03  * ircretaryquit (Remote host closed the connection)
01:29:09  * ircretaryjoined
01:40:32  * skabbesjoined
01:42:45  * dapquit (Quit: Leaving.)
02:03:28  * bradleymeckquit (Quit: Leaving)
02:03:42  * bradleymeckjoined
02:16:53  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
02:20:36  * mikealquit (Quit: Leaving.)
02:23:34  * mikealjoined
02:25:10  * mikealquit (Client Quit)
02:37:01  * TooTallNatequit (Quit: Linkinus - http://linkinus.com)
02:44:45  * bradleymeckquit (Quit: Leaving)
02:46:06  * isaacsquit (Remote host closed the connection)
03:41:45  * travis-cijoined
03:41:46  <travis-ci>[travis-ci] joyent/node#234 (v0.6 - 766f609 : Guillermo Rauch): The build was broken.
03:41:46  <travis-ci>[travis-ci] Change view : https://github.com/joyent/node/compare/bd9fa2e...766f609
03:41:46  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/node/builds/515170
03:41:46  * travis-cipart
03:55:27  * skabbesquit (Quit: skabbes)
04:07:29  * sh1mmerquit (Quit: sh1mmer)
04:08:36  * sh1mmerjoined
04:40:44  * bradleymeckjoined
05:58:52  * isaacsjoined
06:23:26  * isaacsquit (Remote host closed the connection)
06:35:18  * einarosquit (Read error: Operation timed out)
06:42:29  * mikealjoined
06:51:28  * paddybyersquit (Quit: paddybyers)
07:09:19  * paddybyersjoined
07:15:59  * mikealquit (Quit: Leaving.)
07:51:51  * slaskisjoined
08:01:57  * einarosjoined
09:01:00  * cpollack__joined
09:01:20  <indutny>that commit message is too large for me
09:01:20  <cpollack__>That's what she said!!
09:01:24  <indutny>hehehe
09:04:16  * cpollack__quit (Remote host closed the connection)
09:06:57  * cpollackjoined
09:07:13  <indutny>Can you make it harder?
09:07:13  <cpollack>That's what she said!!
09:20:44  * cpollackquit (Remote host closed the connection)
09:49:19  * bradleymeckquit (Ping timeout: 240 seconds)
10:39:12  * einarosquit (Quit: Lost terminal)
11:03:27  * paddybyersquit (Read error: Connection reset by peer)
11:20:16  * piscisaureusjoined
11:20:22  * piscisaureuschanged nick to piscisaureus_
11:20:25  * piscisaureus_quit (Client Quit)
11:22:05  * piscisaureus_joined
11:22:21  * paddybyersjoined
11:26:11  * slaskisquit (Quit: slaskis)
11:38:45  * Raynosquit (Max SendQ exceeded)
11:38:52  * Raynosjoined
12:33:11  * bnoordhuisjoined
13:03:29  * piscisaureus__joined
13:03:29  * piscisaureus_quit (Read error: Connection reset by peer)
13:03:53  * piscisaureus__changed nick to piscisaureus_
13:13:22  <indutny>ryah_: yt? :)
13:52:30  <bnoordhuis>indutny: it's your birthday today?
13:54:15  <indutny>bnoordhuis: yes
13:54:30  <bnoordhuis>indutny: in that case congratulations :)
13:54:43  <indutny>thank you! :)
13:55:18  <indutny>http://img543.imageshack.us/img543/6241/tumblrl8c5jpo1vz1qzhms4.jpg
14:05:08  <piscisaureus_>bnoordhuis: you slacking?
14:05:35  <piscisaureus_>bnoordhuis: also, do you have an opinion on http://groups.google.com/group/nodejs/browse_thread/thread/bd147ec417280a5b
14:08:26  <bnoordhuis>piscisaureus_: bleh, not quite slacking
14:08:44  <bnoordhuis>listenFD? it's gone and you can't reasonably implement it cross-platform
14:09:01  <piscisaureus_>bnoordhuis: how should this guy do what he wants?
14:09:33  <piscisaureus_>bnoordhuis: he's using this daemon manager that gives a process a fourth FD on startup that is is supposed to listen() on
14:09:40  <bnoordhuis>yeah, systemd
14:10:12  <bnoordhuis>there's no good non-hacky solution for that right now
14:11:55  <mmalecki>indutny: happy birthday, sir!
14:12:36  <piscisaureus_>fedor: happy birthday
14:15:34  <indutny>mmalecki: piscisaureus_ thank you!
14:15:55  <piscisaureus_>indutny: re the eio issue
14:16:40  <indutny>piscisaureus_: what `re` means?
14:16:43  <indutny>remind?
14:16:59  <piscisaureus_>indutny: in reply to (I didn't add the actual reply yet :-))
14:17:10  <piscisaureus_>indutny: no matter how events are dispatched, if you consequently initiate more than 1 event from a callback, shit will get out of hand
14:17:48  <piscisaureus_>indutny: but you may want to check whether that 10 events limit is still needed at all
14:18:01  <piscisaureus_>indutny: there's test-eio-race and test-eio-race-2 to try
14:18:08  <indutny>piscisaureus_: I tried, thy works fine
14:18:13  <indutny>piscisaureus_: w/o that limit
14:18:21  <indutny>piscisaureus_: actually, limiting count sounds awful too me
14:18:29  <indutny>piscisaureus_: better limit time, or not limit at all
14:18:41  <piscisaureus_>indutny: time limiting sounds kind of awful to me :-)
14:19:00  <piscisaureus_>indutny: but you should remember that all kinds of limits apply, implicitly or explicitly
14:19:11  <piscisaureus_>it's also not quite easy to reason about
14:19:42  <piscisaureus_>indutny: but about the eio race condition, I think it actually could be an issue that arose due to a bug in darwin
14:19:56  <indutny>piscisaureus_: do you think so?
14:20:01  <piscisaureus_>for which we later added a proper workaround
14:20:16  <indutny>piscisaureus_: got it
14:20:22  <indutny>piscisaureus_: if so, lets remove it!
14:20:31  <indutny>piscisaureus_: are we doing that in unstable branch or not? :)
14:21:03  <bnoordhuis>test-eio-race could use some elucidating comments, i'm not sure what it's trying to prove
14:21:55  <piscisaureus_>indutny: I think you should try to remove it and make sure the eio race tests pass always
14:22:06  <piscisaureus_>indutny: test on linux and mac os
14:22:14  <indutny>piscisaureus_: ok, I'll do
14:22:29  <piscisaureus_>indutny: there is also test-eio-race-4.js
14:22:42  <indutny>piscisaureus_: yeah, I seen
14:23:00  <piscisaureus_>indutny: you should probably stress test a little, e.g. run the test many times and see if you can get it to fail because we want to be really certain about it
14:24:14  <indutny>kk
14:24:29  <indutny>btw, someone on ITerm2 here?
14:24:33  <piscisaureus_>not me
14:24:37  <piscisaureus_>I am on cmd.exe :-p
14:24:48  <txdv>lol ruben ^^
14:25:00  <piscisaureus_>ruben?
14:25:39  <piscisaureus_>txdv: ^
14:26:27  <txdv>i know him, he used to develop with C#, met him at fosdem
14:26:37  <txdv>in the groups.google.com issue
14:26:39  <indutny>ohhh I've something to show you
14:27:02  * cpollackjoined
14:27:08  <indutny>say anything!
14:27:08  <cpollack>That's what she said!!
14:29:34  <indutny>dfffff
14:32:23  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
14:32:36  <bnoordhuis>indutny: so what did you find?
14:32:37  <cpollack>That's what she said!!
14:33:46  * cpollackquit (Remote host closed the connection)
14:33:54  <indutny>bnoordhuis: about what?
14:34:14  <indutny>I've written twss irc bot :)
14:34:19  <indutny>10 lines of code
14:34:24  <bnoordhuis>indutny: i mean the eio limit :)
14:34:29  <indutny>aah
14:34:30  * piscisaureus_joined
14:34:41  <indutny>bnoordhuis: https://github.com/joyent/node/pull/2508
14:35:04  <indutny>bnoordhuis: I found that sometimes at one eio poll more than 10 reqs can be created
14:35:27  <indutny>bnoordhuis: and if that happens, for some reason, first ones won't be processed
14:35:35  <bnoordhuis>indutny: have you tested what happens if you remove the limit?
14:35:49  <indutny>bnoordhuis: yes, destructors are called, callbacks are called
14:35:58  <indutny>bnoordhuis: everything works fine and not leaks memory
14:36:19  <mmalecki>indutny: I've twss as a service :)
14:36:29  <mmalecki>http://twss.nodejitsu.com/api/twss?q=nice%20weather
14:36:29  <bnoordhuis>so why did we put in that limit in the first place?
14:36:38  <mmalecki>http://twss.nodejitsu.com/api/twss?q=it's%20so%20big
14:37:11  <indutny>bnoordhuis: I think because of some eio oddities
14:37:23  <indutny>bnoordhuis: I'm not sure, that's why I was trying to reach out ryah
14:37:30  <indutny>that commit belongs to him
14:37:50  <piscisaureus_>bnoordhuis: I think it was some racy-ness in mac os
14:38:09  <piscisaureus_>bnoordhuis: but eventually I think we just put mutexes around it in eio
14:38:24  <indutny>test was added in 2009
14:38:39  <bnoordhuis>an eternity
14:38:47  <indutny>0d7e88a4
14:38:56  <indutny>finally I found source commit
14:39:13  <piscisaureus_>https://github.com/joyent/libuv/blob/master/src/unix/eio/eio.c#L402-410
14:39:58  <piscisaureus_>indutny: ah maybe we should just talk to felix
14:40:08  <indutny>piscisaureus_: probably!
14:40:16  <indutny>piscisaureus_: is he around ?:)
14:43:10  <indutny>piscisaureus_: apparently, eio-race4 is not related to that eio limit , see 53413598
14:43:34  <bnoordhuis>i remember that thread but what does the mutex have to do with the request limit?
14:43:42  <piscisaureus_>indutny: what is your email address?
14:43:42  <bnoordhuis>is it to prevent long stalls?
14:43:51  <indutny>piscisaureus_: [email protected] or [email protected]
14:45:23  * mmaleckichanged nick to mmalecki[slides]
14:45:42  <indutny>brb, need to restart irssi
14:45:46  * indutnyquit (Quit: leaving)
14:45:53  * indutnyjoined
14:46:09  <indutny>can somebody ping me?
14:46:16  <indutny>I'm testing new colorscheme for irssi
14:46:47  * AndreasMadsenjoined
14:48:20  <txdv>indutny: I get that 11 error in pure libuv
14:48:26  <txdv>while testing a feature that I added
14:48:47  <indutny>oooh
14:48:53  <indutny>color scheme doesn't works :(
14:49:00  <indutny>txdv: hm...
14:49:37  <txdv>`read_watcher` failed: exit code 11
14:49:39  <txdv>oO
14:52:23  <txdv>it is the exit code of the process
14:52:38  <indutny>yeah, I understand
14:53:34  <txdv>so I can use that number to get more specific information
14:55:13  <bnoordhuis>txdv: exit code 11 means that the process segfaulted
14:55:35  <bnoordhuis>set `ulimit -c unlimited` to get a core file you can debug
14:56:56  <txdv>ok, where should the core appear?
14:57:01  <txdv>im using the test-runner
14:58:27  <bnoordhuis>txdv: in $PWD
15:02:05  <indutny>c++ just can't handle what I want it to do
15:04:05  <txdv>i hate c, the hardest part is always compiling
15:04:36  <txdv>./read-watcher: error while loading shared libraries: libuv.so: cannot open shared object file: No such file or directory
15:04:42  <txdv>while libuv.so is in the same directory
15:06:31  <txdv>bnoordhuis: well it doesn't generate core dumps
15:06:47  <txdv>and the other strange problem is that if I compile that test in a stand alone C test case, it runs perfectly
15:06:55  <txdv>in a stand alone C programm
15:08:14  <bnoordhuis>txdv: export LD_LIBRARY_PATH=$PWD
15:25:05  <piscisaureus_>ok I am kind of away - company meeting
15:29:26  <indutny>bnoordhuis: trying to figure out why everything gets delayed with eio max reqs limit
15:29:56  <indutny>bnoordhuis: if every req created insided eio_poll goes at the end of queue - it should work fine
15:30:19  <indutny>bnoordhuis: requests will be processed in order of creation
15:32:53  <indutny>bnoordhuis: and nothing is wrong if so
15:33:25  <txdv>its not dumping cores
15:33:30  <txdv>:(
15:33:45  <txdv>an i hate sometimes hate this entire compilation stuff
15:33:54  <txdv>gdb doesnt track all symbols
15:44:07  <bnoordhuis>indutny: i think the idea of a max req limit is to prevent long stalls
15:44:26  <bnoordhuis>imagine a script that queues 10,000 fs requests in one go
15:46:05  <bnoordhuis>not sure how much of a problem that really is since the bulk of the work takes place in the thread pool
15:47:26  <indutny>bnoordhuis: yeah
15:47:42  <indutny>bnoordhuis: so it just won't let libev to execute it's callbacks
15:47:51  <indutny>bnoordhuis: right?
15:54:22  * isaacsjoined
16:36:15  * kohaijoined
16:37:33  <bnoordhuis>indutny: looking at it again, i'm not sure if 10K requests are really a problem, libuv will wake up the event loop for every request that finishes
16:39:04  <bnoordhuis>iow, i don't know why that 10 req limit was put there...
16:39:27  <txdv>arggh
16:49:17  * einarosjoined
16:54:24  * ljacksonjoined
16:57:48  <indutny>bnoordhuis: just beaten C++ compilation black magic :D
16:58:02  <indutny>bnoordhuis: just two days of work
16:58:10  <indutny>bnoordhuis: probing different variants
17:01:00  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
17:06:56  <indutny>isaacs: hey there
17:07:04  <isaacs>yo
17:07:10  <indutny>isaacs: we're doing hard work
17:07:14  <isaacs>oh, ok
17:07:15  <indutny>isaacs: almost finished it
17:07:22  <indutny>isaacs: this patch is incorrect
17:07:22  <isaacs>:D
17:07:25  <isaacs>i see.
17:07:31  <indutny>isaacs: you can't :)
17:07:37  <indutny>I haven't uploaded it yet
17:07:39  <isaacs>hahah
17:07:44  <indutny>so process is following
17:07:53  <indutny>we'll patch bleeding_edge again
17:07:55  <isaacs>i mean, "i see" in teh sense of "i understand what you have said, and accept it"
17:07:59  <indutny>merge two patches into one
17:08:04  <indutny>and then backport them to 3.6
17:08:08  <isaacs>any ETA?
17:08:15  <indutny>better ask ErikCorryV8 about that
17:08:19  <isaacs>(tomorrow, next week, next month...)
17:09:09  <indutny>isaacs: next week probably, v8 guys ain't accepting patches at weekend
17:10:03  <isaacs>right
17:12:29  * ErikCorryV8quit (Ping timeout: 258 seconds)
17:12:38  <indutny>confirmation ^ :D
17:21:51  * dapjoined
17:28:51  <isaacs>heh
17:29:00  <isaacs>well, friday is almost over in Europe, right?
17:29:04  <isaacs>isn't it like 5pm there?
17:38:16  <indutny>where?
17:38:24  <indutny>it's like 0:38 am here
17:38:28  <indutny>oh
17:38:29  <indutny>12:38 am
17:38:30  <indutny>sorry
17:44:19  <indutny>isaacs: follow that
17:44:26  <indutny>http://codereview.chromium.org/9212006
17:46:53  <isaacs>indutny: where are you, anyway?
17:46:57  <isaacs>asia somewhere?
17:47:24  <indutny>isaacs: Russia, Omsk
17:47:34  <indutny>isaacs: GMT+7
17:47:45  <indutny>it's near to Novosibirsk
17:48:01  <indutny>Asia
17:48:51  <isaacs>yikes
17:48:59  * sh1mmerquit (Quit: sh1mmer)
17:49:48  <isaacs>that's awesome. with you in omsk, and koichi in japan, nodejs has 24 hour core team activity :)
17:49:48  <indutny>isaacs: I'm not asian if you're curious :)
17:49:53  <indutny>hahahaha
17:49:57  <indutny>yes
17:50:00  <indutny>24/7
17:50:15  <isaacs>indutny: yeah, i knew you were in russia somewhere, but the timezone put you way east
17:50:24  <indutny>that's it
17:50:32  <isaacs>i guess russia is just really really big.
17:51:36  <isaacs>stretches from europe to north america
17:54:52  <indutny>yeah, Alaska was ours until 1867
17:55:00  <indutny>so we was in North america too :)
17:55:41  <AvianFlu>if we hadn't paid you 2 cents an acre back then, substack and jesusabdullah would be speaking russian!
17:55:54  <indutny>AvianFlu: ahhahaha :)
17:56:04  <indutny>AvianFlu: right, that was cheap deal
17:58:12  * mikealjoined
18:04:10  * dshaw_quit (Quit: Leaving.)
18:07:58  <indutny>isaacs: oooh, that patch was incorrect
18:08:18  <indutny>isaacs: doing reinterpret_cast is always bad idea
18:08:30  * dshaw_1joined
18:13:38  * TooTallNatejoined
18:13:43  <bnoordhuis>https://github.com/joyent/node/issues/2474#issuecomment-3479265 <- http proxy/tunneling support, aye or nay?
18:34:53  * mikealquit (Quit: Leaving.)
18:35:32  <isaacs>bnoordhuis: this is kind of a pita for npm right now.
18:35:49  <bnoordhuis>isaacs: how so?
18:35:54  <isaacs>bnoordhuis: i think it'd be worth *trying* to do it in userland as part of request (if no one already has tried this) and see what the blockers are.
18:36:10  <isaacs>bnoordhuis: well, npm has this "proxy" config you can use. but if your proxy is http and the registry is https, then you're out of luck.
18:36:12  <isaacs>it just dies very badly.
18:36:18  <bnoordhuis>right
18:36:54  <isaacs>because request doesn't do any kind of tunneling, it just juggles the Authorization vs ProxyAuthorization headers, and puts the full registry url in the request path.
18:37:01  <isaacs>ie, basic simple http proxying.
18:37:18  <isaacs>i haven't looked into the issue deeply.
18:37:45  <isaacs>is there something preventing doing this outside of node-core?
18:37:52  <isaacs>if so, what's the minimal change we could make to unblock it?
18:38:03  <isaacs>if not, it should be a fix to request, imo.
18:38:04  <bnoordhuis>not since a couple of patches to master have landed
18:38:42  <isaacs>yes, http is first-class, and http->https proxying is very obnoxious. but if we can get away with it, i'd prefer to do it in request first, experiment a bit, see what works and what doesn't, and then have a conversation about which bits to pull into core.
18:38:54  <isaacs>the API for proxying that request exposes IS really nice.
18:39:06  <isaacs>you just do { "proxy": "http://[email protected]:1234" }
18:39:18  <isaacs>in teh config object
18:39:40  <isaacs>maybe we could eventually pull that kind of API into node-core for http.request() and https.request()
18:39:46  <bnoordhuis>sounds reasonable
18:40:13  <isaacs>we might get it done in request, and then go, "oh, actually, if you want this, just use this userland lib"
18:40:24  <isaacs>most people doing a lot of http client work use request already anyhow
18:42:20  * `3rdEdenjoined
18:43:50  * mikealjoined
18:50:27  * mikealquit (Quit: Leaving.)
19:09:16  * igorzijoined
19:11:51  * brsonjoined
19:24:22  <AndreasMadsen>bnoordhuis: #2463 this is now fixed, tested and check - lets land this.
19:34:43  * pieternjoined
19:37:09  <isaacs>AndreasMadsen, bnoordhuis: it looks like https://github.com/AndreasMadsen/node-1/commit/0f61378b81239988c6c8647b0f13143b4ff5b1e9 adds api
19:37:57  <AndreasMadsen>isaacs: That is internal API
19:38:10  <isaacs>hrm... why do we need this?
19:38:16  <isaacs>(sorry, coming in halfway thorugh, i know)
19:39:22  <AndreasMadsen>isaacs: You remember the very long discussion about logging output from fork ...
19:39:29  <isaacs>+1 on the rest of it
19:39:55  <isaacs>AndreasMadsen: right. so... it should go in master.
19:40:28  <AndreasMadsen>isaacs: well the conclusion was if this was really required people should use the internal stdoutStream API
19:40:29  <isaacs>i'm not opposed to it in principle. just that we have to be very strict about the changes we add in 0.6
19:40:49  <isaacs>if it's really required, they should use spawn
19:40:59  <AndreasMadsen>isaacs: But that was not possible because of this two missing lines
19:41:50  <isaacs>we should not change the exposed api, even things that are labeled "internal"
19:41:55  <AndreasMadsen>isaacs: The argument against using spawn was that it would require to use both stdin and stdout, meaning that stdout could not be logged.
19:42:22  <isaacs>ok, so log from stderr, or use spawn and pipe the stdout and stdin streams manually
19:42:50  <AndreasMadsen>isaacs: Also the problem as already been fixed in master by adding an {silent: options}
19:42:53  <isaacs>child.stdout.pipe(logger).pipe(process.stdout)
19:43:56  <AndreasMadsen>isaacs: The stdout stream is used to send IPC data from child to parent, this should not be logged.
19:44:19  <isaacs>AndreasMadsen: what do you mean by "logged"?
19:44:25  <isaacs>you mean, written to the paren'ts stdout stream?
19:44:40  <AndreasMadsen>isaacs: I'm was never a part of the discussion I'm just repeating arguments
19:44:46  <isaacs>yeah
19:44:53  <AndreasMadsen>isaacs: As in logged to a file
19:44:59  <AndreasMadsen>or database
19:44:59  <isaacs>ok...
19:45:04  <isaacs>so... don't log it?
19:45:06  <isaacs>i don't get it.
19:45:29  <AndreasMadsen>What do you not get
19:45:47  <isaacs>i don't get what's not possible without adding internal api, and then telling people to use the internal undocumented api to do what they want.
19:46:04  <isaacs>the problem is, even though it's arguably a good change, we're just adding roadblocks that prevent people from upgrading in the future.
19:47:12  <AndreasMadsen>isaacs: fork provides a two way communication channel though stdin without polluting stdin
19:47:14  <isaacs>because, what's going to happen is, when they want to upgrade to 0.7 or 0.8, it'll be a much bigger hassle. or worse, if in 0.6.9, we find some issue, and change this api thiking it's safe because it's internal, now the "stable" branch isn't stable any more.
19:48:18  <AndreasMadsen>isaacs: The core issue is that it is NOT possible to log both stdout and stderr, while have a two way communication channel.
19:48:44  <isaacs>AndreasMadsen: why not just have the parent start a net server on a file socket, and then the children connect to it?
19:49:13  <AndreasMadsen>isaacs: The arguments in the issue was that this was possible using internal API, but it tuned stdoutStream and stderrStream was missing.
19:49:16  <isaacs>AndreasMadsen: or just say, sorry, you can't do that in 0.6.
19:49:32  <isaacs>AndreasMadsen: i didn't make that argument in the issue.
19:49:36  <isaacs>that argument is invalid.
19:49:50  <AndreasMadsen>indeed
19:49:53  <isaacs>internal api's should not be used outside of node-core.
19:49:58  <isaacs>it causes huge problems.
19:50:11  <isaacs>i've been the offender with npm, many, many, many times.
19:50:13  <isaacs>:)
19:50:38  <isaacs>but like, from 0.1 on up, every time someone uses an internal api, it's a huge pita later, because we change it.
19:50:57  <isaacs>and then their stuff breaks, and they're like, "But you said this was the stable branch!"
19:51:01  <isaacs>(and they're right!)
19:51:32  <isaacs>if you use an internal API, it should be with the understanding that it will disappear at any moment, and there will be no replacement.
19:51:53  <isaacs>and no one actually can live up to that, even though we think we can.
19:51:56  <AndreasMadsen>I totally agree
19:52:02  <isaacs>anyway... rant rant rant...
19:52:23  <isaacs>the rest of the changes are great. lgtm.
19:52:48  <isaacs>but this isn't enough justification to add stderrStream and stdoutStream
19:53:15  <AndreasMadsen>Anyway I **think** bnoordhuis made the invalid, so I don't know
19:53:17  <isaacs>i'd be +1 on adding it to master, along with docs and tests, and promoting it to a proper api, if we think it's the best way to address the issue.
19:53:23  <isaacs>haha
19:53:40  <AndreasMadsen>isaacs: why docs and tests
19:53:55  <AndreasMadsen>Nothing public has been aded
19:54:20  <isaacs>AndreasMadsen: i mean, if we are going to add stderrStream and stdoutStream, and suggest people use them, then let's promote it from internal to supported.
19:54:35  <isaacs>if it's actually the best api for the problem, and it's a feature we want to add.
19:55:38  <AndreasMadsen>We should not suggest people to use use std*Streams there is allready a stdin, stderr, and stdout channel in fork when using a silent option in node 0.7, solving all issues
19:55:55  <indutny>bnoordhuis: yt?
19:56:03  <indutny>eio is most dumbest thing I've ever seen
19:56:08  <indutny>:D
19:56:10  <AndreasMadsen>As said I do not know why that invalid argument was made
19:56:28  <indutny>bnoordhuis: they're are using FILO queue
19:56:37  <indutny>bnoordhuis: that's why req limit isn't working
19:57:30  * mikealjoined
19:58:27  <isaacs>AndreasMadsen: ok, if you say "no" to adding std*Stream options in 0.7, then i'm sold :)
19:59:09  <AndreasMadsen>isaacs: I most definitely say NO, sold on what?
19:59:29  <isaacs>AndreasMadsen: you've sold me. i agree with you. "no" it is :)
19:59:48  <isaacs>AndreasMadsen: "no" should be our default anyway
20:00:23  <AndreasMadsen>isaacs: Then you haven't developed a single module to cluster, have you
20:00:28  <AndreasMadsen>:)
20:00:59  <isaacs>AndreasMadsen: well, the cluster stuff needed to either get better or go away. going away wasn't an option in this case, so it's good that you've done a bunch of work on that.
20:01:15  <isaacs>but of course, that can't can't be in 0.6, because it's adding stuff.
20:01:30  <AndreasMadsen>And it is fine with me
20:01:45  <isaacs>i should clarify: the default answer for master is "hm... i don't know... do we really need this?" the default answer for 0.6 is "absolutely not"
20:03:21  <AndreasMadsen>isaacs: It was ryan there made the invalid argument :)
20:03:30  <isaacs>ah, hahah
20:04:03  <isaacs>which issue is that?
20:04:04  <isaacs>link?
20:04:07  <AndreasMadsen>2012-01-06 20:41:47 #libuv IRC
20:04:14  <AndreasMadsen>http://piscisaureus.no.de/log/2012-01-06#20:41:47.968
20:04:17  <indutny>bnoordhuis: oh, or not
20:04:27  <indutny>bnoordhuis: missed that they have qs and qe
20:05:08  <indutny>bnoordhuis: nope, they're using FIFO
20:05:27  <isaacs>AndreasMadsen: Oh!!
20:05:32  <isaacs>i see.
20:05:34  <isaacs>it's actually not adding api.
20:05:37  <isaacs>it's just adding a default value.
20:05:44  <isaacs>which otherwise would be "undefined" rather than "null"
20:05:53  <AndreasMadsen>Yes
20:05:57  <isaacs>AndreasMadsen: ok.
20:06:04  <isaacs>then your patch +1 lgtm etc etc
20:06:16  <AndreasMadsen>awesome
20:06:19  <isaacs>yes, it should get a default value, and that doesn't add api. i was misreading it.
20:06:33  <isaacs>and i need to smack ryah_ for suggesting that people use internal apis
20:07:03  <AndreasMadsen>awesome: yes please do so
20:07:26  <AndreasMadsen>especially internal API there do not work :)
20:10:23  <AndreasMadsen>to all: Please remind ryah about https://github.com/joyent/node/pull/2470 - he asked for it 5h after the previously cluster patch was landed and now he has forgot it.
20:10:50  <AndreasMadsen>note he got it 3 minutes later :)
20:11:27  <indutny>bnoordhuis: ok, now I understand why that madness happens
20:11:55  * isaacschanged nick to isaacs_lunch
20:12:19  * isaacs_lunchquit (Remote host closed the connection)
20:13:59  <indutny>bnoordhuis: please ping me once you'll be back
20:41:54  * AndreasMadsenquit (Remote host closed the connection)
20:46:01  * AndreasMadsenjoined
21:01:36  * mikealquit (Quit: Leaving.)
21:13:29  * indutny_joined
21:15:08  * indutnyquit (Ping timeout: 268 seconds)
21:27:08  * mikealjoined
21:27:09  * indutny_changed nick to indutny
21:32:24  * mikealquit (Quit: Leaving.)
21:47:34  * mikealjoined
21:50:38  <dap>isaacs: https://skitch.com/dapsays/g2wq2/rev8-10min
21:50:52  <dap>gotta run; will be exploring this and the slightly more complex example a bit later.
21:51:55  <dap>oops, he's not even here.
22:08:50  * mikealquit (Quit: Leaving.)
22:11:05  * mikealjoined
22:11:27  * mikealquit (Client Quit)
22:13:16  * mikealjoined
22:16:26  * mikealquit (Client Quit)
22:17:24  * mikealjoined
22:35:53  <bnoordhuis>indutny: you rang?
22:37:51  * sh1mmerjoined
22:40:33  <indutny>bnoordhuis: sorry , sleeping now :)
22:40:44  <bnoordhuis>indutny: hah okay, sleep tight
22:40:55  <bnoordhuis>ping me tomorrow if you want
22:41:02  <indutny>ok
22:41:03  <indutny>ttyl
22:43:57  * sh1mmerpart
22:44:06  * sh1mmerjoined
22:45:16  * dshaw_1changed nick to dshaw_
22:47:52  * sh1mmer_joined
22:48:24  * sh1mmer_quit (Client Quit)
22:50:00  * sh1mmerquit (Remote host closed the connection)
22:50:32  * sh1mmerjoined
22:51:52  * mmalecki[slides]changed nick to mmalecki
23:01:04  * AndreasMadsenquit (Remote host closed the connection)
23:07:47  * `3rdEdenquit (Quit: Zzz)
23:16:16  * dapquit (Quit: Leaving.)
23:16:40  * dapjoined
23:21:15  * mralephjoined
23:29:31  * mikealquit (Quit: Leaving.)
23:32:21  * Raynos_joined
23:32:32  * Raynosquit (Excess Flood)
23:32:33  * Raynos_changed nick to Raynos
23:39:49  <mmalecki>any chance something got fucked up lately? we've many users complaining about node unable to write to /tmp with EUNKNOWN. they are able to put files there by hand.
23:40:31  <mmalecki>not sure if it's node or npm tho (EUKNOWN is an issue anyway).
23:40:33  * paddybyers_joined
23:40:39  <mmalecki>and no, can't repro.
23:40:59  <rmustacc>What OS are they using?
23:41:12  <rmustacc>Can they reproduce it?
23:41:26  <mmalecki>OS X, and yeah, they can
23:41:27  * mralephquit (Quit: Leaving.)
23:41:51  <rmustacc>Then get them to use dtrace to figure out what system call is actually setting errno.
23:42:02  <bnoordhuis>or truss
23:42:10  <rmustacc>Do they have truss on OS X?
23:42:25  <bnoordhuis>it's what i use on os x anyway
23:42:34  <rmustacc>Did you install it?
23:42:43  <bnoordhuis>i think so, it might be a dtrace wrapper
23:42:44  <mmalecki>that might work. `dtrace jitsu deploy`? (`jitsu` is the app it happens in)
23:43:15  <rmustacc>Shame Mac OS X doesn't have pgrep.
23:43:18  * txdv_joined
23:43:46  <rmustacc>Probably just use dtruss
23:43:52  <rmustacc>Probably the easeir way to do it.
23:43:53  <mmalecki>rmustacc: what's the invocation?
23:44:04  <rmustacc>man dtruss
23:44:04  <mmalecki>dtruss jitsu deploy?
23:44:26  <mmalecki>I mean, I know. wondering what info you need.
23:44:49  <bnoordhuis>the failing syscall and its errno
23:45:03  <bnoordhuis>but preferably the whole log
23:45:29  <mmalecki>bnoordhuis: so, just `dtruss jitsu deploy`? no options?
23:45:31  <mmalecki>or -a?
23:46:05  <bnoordhuis>just dtruss should be okay, i think
23:47:23  <CIA-115>libuv: Ben Noordhuis v0.6 * rac218a7 / (5 files): test: add lots of refcount tests - http://git.io/AZ_Kgg
23:47:24  <CIA-115>libuv: Ben Noordhuis v0.6 * rdc3b80a / (test/echo-server.c test/task.h test/test-list.h): test: add udp4_echo_server helper - http://git.io/sqtZrw
23:47:25  <CIA-115>libuv: Ben Noordhuis v0.6 * r6d9c9a5 / (5 files in 5 dirs): Add uv_loop_refcount() function, debug tool. - http://git.io/RMEq0Q
23:47:27  <CIA-115>libuv: Ben Noordhuis v0.6 * ra13584b / src/unix/core.c : unix: remove unused #define - http://git.io/9rxb5w
23:47:38  <bnoordhuis>^ lots of new, failing tests :(
23:47:50  <mmalecki>bnoordhuis: you broke stuff. again.
23:48:09  <bnoordhuis>no, but i expose what is broken
23:48:41  * ljacksonquit (*.net *.split)
23:48:41  * paddybyersquit (*.net *.split)
23:48:42  * txdvquit (*.net *.split)
23:48:42  * paddybyers_changed nick to paddybyers
23:49:09  * travis-cijoined
23:49:09  <travis-ci>[travis-ci] joyent/libuv#29 (v0.6 - 6d9c9a5 : Ben Noordhuis): The build is still failing.
23:49:09  <travis-ci>[travis-ci] Change view : https://github.com/joyent/libuv/compare/52fba1a...6d9c9a5
23:49:09  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/libuv/builds/520923
23:49:09  * travis-cipart
23:49:30  <mmalecki>bnoordhuis: TOTALLY YOUR FAULT ^.
23:50:42  <mmalecki>well... dtruss needs sudo and it works with sudo for them
23:50:48  <mmalecki>any hints here?
23:51:33  <mmalecki>hm actually, would `sudo dtruss sudo -u <username> jitsu deploy`?
23:51:38  <mmalecki>*would it work?
23:52:25  <bnoordhuis>you need to trace child process too in that case
23:52:45  <rmustacc>You could just dtruss -n node
23:52:57  <rmustacc>And then tell them to do the deploy again.
23:53:02  <rmustacc>Without the dtruss invocation.
23:53:12  <mmalecki>oh! thanks rmustacc, will try that
23:53:12  <rmustacc>That will truss all node programs then, rather than the one it spawns.
23:53:35  * ljacksonjoined
23:54:29  <igorzi>bnoordhuis: what isolates tests work in node?
23:54:51  <bnoordhuis>igorzi: test-isolates, test-isolates2 and test-isolates-ping-pong
23:54:53  * mikealjoined
23:55:58  * mikealquit (Client Quit)
23:56:39  * mikealjoined
23:57:09  <bnoordhuis>igorzi: https://github.com/joyent/node/commit/13324bf <- what was the reason for that change?
23:59:15  <mmalecki>https://raw.github.com/gist/c7fbc4bf0848245d9ab9/89ef9d4f6a0f6a76ed396ec386b88eb7a6fd0e98/usdtruss2
23:59:52  <igorzi>bnoordhuis: one sec