00:06:48  <dap>isaacs: it turns out that in master, we can build both v8 andnode with "-fno-omit-frame-pointer" without floating a patch or anything. (thanks, gyp!) We definitely want this for sunos; do you have any thoughts on doing it globally vs. just for sunos?
00:07:44  <isaacs>dap: where does that flag have to be?
00:07:48  <isaacs>in v8's gyp file or ours?
00:07:52  <dap>ours
00:08:02  <isaacs>i think gyp can do something like ["OS===solaris", { blah blah blah }]
00:08:09  <dap>yeah, we definitely can.
00:08:44  <isaacs> [ 'OS=="solaris"', {
00:08:44  <isaacs> 'libraries': [
00:08:44  <isaacs> '-lkstat',
00:08:45  <isaacs> ],
00:08:47  <isaacs> }],
00:08:49  <isaacs>that's in our gyp file already
00:09:01  <isaacs>dap: patch welcome :)
00:09:06  <dap>sounds good.
00:09:25  * isaacsputs another small pebble in the "dap is node's sunos/dtrace guy" bucket.
00:09:45  <dap>:)
00:12:01  * AvianFlujoined
00:13:46  <dap>isaacs: would a similar patch for the SConstruct file for v0.6 also be welcome?
00:13:57  <isaacs>sure
00:14:19  <isaacs>the next (and hopefully final) 0.6 version drops on Thursday
00:14:27  <dap>okay.
00:15:16  <TooTallNate>dap: isaacs: that wouldn't make existing native modules already compiled for 0.6 stop working right?
00:15:38  <dap>TooTallNate: I can't see how.
00:15:49  <TooTallNate>ok cool
00:15:57  <dap>It's just turning off an optimization that itself violates the x86 ABI.
00:16:14  <TooTallNate>haha, ok sounds good
00:24:23  * piscisaureus_joined
00:29:50  <isaacs>review, plz?
00:29:51  <isaacs>https://gist.github.com/913e61fa9d0945096373
00:30:17  <isaacs>(mostly see if there's anything major missing from the ChangeLog that you know of)
00:31:59  <piscisaureus_>igorzi: yt?
00:32:18  <igorzi>piscisaureus_: sup?
00:32:29  <piscisaureus_>igorzi: re: I’ll start looking at node/libuv issues and see if there’s anything there that needs to be addressed. Do you know of anything that I can help with?
00:32:45  <piscisaureus_>igorzi: I can think of a man-year of work? Do you prefer thinking or coding?
00:33:19  <bentkus_>o hello!
00:33:52  <igorzi>piscisaureus_: let's skype tomorrow (or today for you).. i need to talk to you about something else as well
00:33:54  <TooTallNate>isaacs: curious, why does NODE_VERSION_IS_RELEASE change?
00:34:25  <isaacs>TooTallNate: when that's set to 0, it puts the -pre on teh version
00:34:30  <isaacs>TooTallNate: when it's 1, it doesn't.
00:34:42  <piscisaureus_>igorzi: Yeah, ok. (is it something exciting?)
00:34:44  <TooTallNate>oh ok, makes perfect sense
00:35:01  <isaacs>release commits set it to 1, then it gets set back to 0 on the "now working on x.y.z" commit immediately after.
00:35:13  <igorzi>piscisaureus_: it's a named-pipe issue that we need to fix
00:35:21  <TooTallNate>right, i never even though about how that was done. pretty clever :)
00:35:37  <piscisaureus_>igorzi: it better not be the lack of shutdown() support :-)
00:35:45  <igorzi>piscisaureus_: it's not :)
00:35:51  <piscisaureus_>right. ok
00:35:58  <igorzi>piscisaureus_: gtg
00:36:03  <piscisaureus_>igorzi: that's good actually, that means I can go to sleep now
00:36:09  <piscisaureus_>igorzi: ttyl
00:41:34  <TooTallNate>isaacs: btw, what does RFH stand for?
00:41:37  <TooTallNate>request for help?
00:42:31  <isaacs>TooTallNate: yes
00:42:37  <isaacs>like RFC or RFP
00:42:44  <isaacs>(comment, proposal)
00:43:33  <TooTallNate>right, that's what i figured, just wasn't sure about the "H" :p
00:47:47  <mjr_>So we are now mere days away from moving some of our node stuff to smartos.
00:48:18  <mjr_>And then we'll be able to busy out some of that hot, hot ustack action.
00:48:27  <mjr_>bust, that is.
00:55:09  <isaacs>windows and linux look good. os x looks good. solaris is timing out on every test.
00:55:15  <isaacs>something's odd with my vm,i think
00:55:45  <isaacs>mjr_: nice
00:57:06  <isaacs>testing on a real smartmachine now
01:05:44  <isaacs>alright, proper SmartOS smartmachine worked fine.
01:05:55  <isaacs>i'll investigate the vmware image (or probably,just trash it and build a new one) later.
01:05:58  <isaacs>dropping this thing
01:06:06  <bnoordhuis>good luck with the release, isaacs
01:06:07  * bnoordhuissigns off
01:06:16  <isaacs>bnoordhuis: it's pretty much done
01:06:18  <isaacs>just pushing live now
01:06:21  <isaacs>email etc
01:06:27  <isaacs>thanks
01:07:48  <CIA-101>node: isaacs master * r349cab0 / Makefile :
01:07:48  <CIA-101>node: Check the version before building tarball
01:07:48  <CIA-101>node: I keep forgetting to do this. It's such a stupid thing. - http://git.io/liox7g
01:07:49  <CIA-101>node: isaacs master * r99059aa / (7 files in 6 dirs): (log message trimmed)
01:07:49  <CIA-101>node: 2012.02.07, Version 0.7.3 (unstable)
01:07:49  <CIA-101>node: * Upgrade V8 to 3.9.2
01:07:49  <CIA-101>node: * Revert support for isolates. (Ben Noordhuis)
01:07:50  <CIA-101>node: * cluster: Cleanup docs, event handling, and process.disconnect (Andreas Madsen)
01:07:51  <CIA-101>node: * gyp_addon: link with node.lib on Windows (Nathan Rajlich)
01:07:51  <CIA-101>node: * http: fix case where http-parser is freed twice (koichik)
01:07:52  <CIA-101>node: isaacs master * re612143 / src/node_version.h : Now working on 0.7.4 - http://git.io/GKE4NA
01:08:07  <benvie>does module.children ever get used?
01:08:15  <benvie>it doesn't in module.js
01:09:22  <benvie>don't think I've ever seen it used
01:10:39  <isaacs>benvie: i dunno
01:10:53  <isaacs>LETS REMOVE IT AND SEE IF IT BREAKS ANYTHING IT PROBABLY IS OK!!
01:11:00  * isaacsoh wait, loudbot isn't in here
01:11:03  * bnoordhuisquit (Ping timeout: 248 seconds)
01:11:05  <isaacs>:)
01:11:08  <benvie>haha that is the best of all options
01:12:54  <benvie>sometimes it's like that game where you're god and you have to balance good and evil. Sometimes things need to be made awesome. Sometimes they need to fucking explode
01:16:46  * dshaw_quit (Read error: Operation timed out)
01:17:39  * dshaw_joined
01:20:11  * mjr_quit (Quit: mjr_)
01:21:20  * sj26quit (Excess Flood)
01:21:42  * travis-cijoined
01:21:42  <travis-ci>[travis-ci] joyent/node#384 (master - e612143 : isaacs): The build is still failing.
01:21:42  <travis-ci>[travis-ci] Change view : https://github.com/joyent/node/compare/986785c...e612143
01:21:42  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/node/builds/641898
01:21:42  * travis-cipart
01:21:50  <isaacs>no surprise there
01:23:58  * sj26joined
01:31:11  * orlandovftwquit (Ping timeout: 245 seconds)
22:57:39  <piscisaureus_>^^^ logging was down for couple of hours ^^^
22:59:26  <mmalecki>anyways, switching channel into party mode
22:59:36  <mmalecki>i'm turning 18 in a minute
23:00:14  <piscisaureus_>So what does that mean in poland?
23:00:39  <mmalecki>piscisaureus_: you can legally buy alcohol
23:00:42  <mmalecki>and that's about it
23:00:53  <piscisaureus_>hmm
23:01:15  <mmalecki>yeah, well, also our law is pretty fucked in some cases
23:01:23  <mmalecki>I just found out
23:01:44  <piscisaureus_>... that you cannot have sex until you're 25?
23:02:01  <mmalecki>no
23:02:07  <piscisaureus_>ok that was not funny
23:02:12  <mmalecki>parents can get me kicked out off school anytime they want
23:02:29  <piscisaureus_>huh
23:02:40  <mmalecki>long story
23:03:04  <piscisaureus_>You mean there is no compulsory education after 18?
23:03:08  <piscisaureus_>that is pretty common
23:03:50  <mmalecki>no, I mean, even if I want to learn, parents can get me kicked out of school
23:04:08  <mmalecki>painful
23:04:12  <piscisaureus_>you mean, you're own parents?
23:04:17  <mmalecki>yeah
23:04:21  <piscisaureus_>hmm, weird
23:04:36  <mmalecki>welcome to poland >.<
23:04:44  <piscisaureus_>here 18 just means independent, in theory you are allowed to do whatever you want
23:04:50  <piscisaureus_>(*) condition apply
23:05:26  <isaacs>in the US you have to be 21 to drink, and it's very hard to rent a car or get very much credit before 25
23:05:36  * paddybyersquit (Quit: paddybyers)
23:05:46  <piscisaureus_>I think here you have to be 21 to drive a heavy motorbike
23:06:00  <piscisaureus_>And yeah some stuff is hard
23:06:25  <mmalecki>humans are hard
23:07:12  * dshaw_quit (Quit: Leaving.)
23:07:36  <piscisaureus_>isaacs: getting very much credit should be hard for everyone.
23:08:38  <piscisaureus_>isaacs: do you understand what mikeal and tj mean with that "composability" stuff
23:08:38  <isaacs>well, the credit you can get, and the price you pay for it, should be based on the likelihood that you'll pay it back.
23:08:41  <isaacs>yes.
23:09:01  <isaacs>they mean, i can hand a request off to one function, that adds a few headers, and then to another function, that adds a few more headers.
23:09:07  <isaacs>so you can compose things
23:09:13  <piscisaureus_>Oh ok
23:09:29  <piscisaureus_>so what do we need to change, other than writeHead?
23:09:44  <isaacs>there are a bunch of relatively minor cleanup things that mikeal wants to do.
23:09:50  <isaacs>he's waiting for a clean `make test1
23:10:12  <isaacs>i'm hearing that a lot, actually. that's why i really want us to make every single test pass as soon as we can.
23:10:16  <isaacs>0.6 is a lost cause, whatever.
23:10:45  <piscisaureus_>isaacs: there's not so many failures right?
23:10:51  <isaacs>after 0.6.11, and this api doc refactor, that's a bit item looming on my agenda.
23:11:01  <isaacs>piscisaureus_: there are so many. >0 = so many :)
23:11:19  <isaacs>piscisaureus_: it's not very bad now. there are like 1-5 failures every time on the mac.
23:11:53  <isaacs>but like, if there's a test that we know fails a lot, we should either fix it, or decide that we're not going to fix it, and remove it.
23:12:04  <isaacs>probably, we should fix the feature that it's testing
23:13:02  <piscisaureus_>isaacs: it is unreasonable for an unstable branch to expect that every test passes every time
23:13:26  <isaacs>sure.
23:13:30  <isaacs>they're going to break and get fixed, and so on
23:13:31  <piscisaureus_>isaacs: but I agree we have to pass 100% on the next stable branch
23:13:37  <piscisaureus_>and keep it that way
23:13:57  <isaacs>but it's also unreasonable to have a test that we all know fails, and go "oh, that one, yeah, whatever, don't worry about that one"
23:14:10  <isaacs>like, we need to either fix it, mark it as "todo" or something
23:14:11  <piscisaureus_>Actually, maybe it's not unreasonable to have 100% pass on unstable
23:14:17  <piscisaureus_>yeah
23:14:39  <isaacs>we could put the failures in a special folder, and say "all these must pass before moving to stable branch mode"
23:14:56  <isaacs>and maybe the diff between stable and unstable is "there's nothing in test/todo/*.js"
23:15:11  <piscisaureus_>isaacs: we should have a command to run only the failures
23:15:20  <isaacs>right
23:15:20  <piscisaureus_>like, vcbuild test-broken
23:15:29  <isaacs>and maybe move any passing ones into test/simple or something
23:15:39  <isaacs>or at least list out which ones passed so you can do this manually
23:15:52  <piscisaureus_>hmm
23:16:01  <piscisaureus_>it would be good to push failures on a list or something
23:16:18  <piscisaureus_>because now when I want to fix a test, I have to run make test-all
23:16:20  <piscisaureus_>which takes long
23:16:25  <piscisaureus_>which is very demotivating
23:16:41  <piscisaureus_>test-debugger-repl needs fixing
23:16:52  <isaacs>yeah
23:17:09  <isaacs>test-fs-chmod fails on my laptop, but only when run as part of `make test`
23:17:14  <isaacs>not when i run it manually
23:17:17  <isaacs>i don't get it
23:17:24  <piscisaureus_>dgram-ping-pong also intermittently fails for me
23:17:27  <piscisaureus_>I think everywhere
23:17:58  <piscisaureus_>isaacs: what if you run `node test/simple/test-fs-chmod | cat`
23:18:00  <mmalecki>this message passing thing, fork2 is easy to fix
23:18:21  <mmalecki>we kill worker too early there
23:18:22  * creationixjoined
23:18:52  <mmalecki>and if load is too big (as when running tests) it fails
23:18:53  * dshaw_joined
23:22:21  <creationix>can someone help me with "UNKNOWN, unknown error" in libuv?
23:22:27  <creationix>this time I can't make sense of the strace
23:23:41  <creationix>https://gist.github.com/44c69c503c490a080493
23:24:19  <creationix>on line 20 I open /dev/input/js0 and log the file descriptor 5
23:24:33  <creationix>the next thing I do is read the first 8 bytes in that file
23:24:49  <creationix>in the after_fs callback, I have a UNKNOWN uv error
23:25:08  <creationix>I don't see it ever issuing a read system call of any kind
23:26:46  <isaacs>piscisaureus_: that works, too
23:28:25  <creationix>hmm, actually I can't find a system call for the open call
23:30:10  <piscisaureus_>creationix: does that happen consistently?
23:31:03  * sh1mmerjoined
23:31:05  <piscisaureus_>creationix: try to put a breakpoint in uv_translate_sys_error
23:34:24  <creationix>piscisaureus_, how do I set a breakpoint?
23:34:54  <piscisaureus_>break uv_translate_sys_error
23:34:58  <piscisaureus_>in gdb
23:36:37  <creationix>the errno is 29
23:36:40  <piscisaureus_>aha
23:36:44  <piscisaureus_>now open errno.h :-)
23:37:18  <tjfontaine>espipe, nice
23:37:37  <piscisaureus_>on linux, that would be ESPIPE yeah
23:37:52  <piscisaureus_>probably treating a pipe or socket like a file
23:38:07  <piscisaureus_>possibly one of the stdio handles
23:38:10  <creationix>the file is /dev/input/js0
23:38:14  <creationix>a device
23:38:24  <piscisaureus_>creationix: how do you open it?
23:38:41  <creationix>"r" mode and 0644 permissions
23:38:44  <creationix>fs.open
23:38:47  <piscisaureus_>aha
23:39:04  <piscisaureus_>creationix: file streams use pwrite and pread to read
23:39:21  <piscisaureus_>which means they read from a specified offset which doesnt work for this device
23:39:37  <creationix>except I'm passing in null
23:39:41  <creationix>as the offset
23:39:46  <piscisaureus_>creationix: doesn't matter
23:40:07  <TooTallNate>bnoordhuis: so did buffertools work on windows?
23:40:44  <creationix>piscisaureus_, fwiw, it works fine in node https://github.com/nodebits/linux-joystick/blob/master/read-loop.js
23:40:47  <creationix>but not in luvit
23:41:06  <piscisaureus_>weard-o
23:42:01  <creationix>actually, I'm passing in the int 0 for offset
23:42:05  <piscisaureus_>creationix: try to pass -1 as offset
23:42:06  <creationix>but that's the same as NULL in C right?
23:42:12  <piscisaureus_>creationix: yes
23:42:29  <piscisaureus_>creationix: but we translate javascript null to -1 in this case
23:42:44  <creationix>ahh, much better
23:42:50  <piscisaureus_>because when the offset is 0 then the user probably wants to read at offset 0 :-)
23:42:58  <creationix>right
23:43:09  <creationix>thanks
23:43:14  <piscisaureus_>creationix: https://github.com/joyent/libuv/blob/master/src/unix/fs.c#L234-236
23:43:27  <piscisaureus_>creationix: btw - I would expect that device to be pollable
23:43:38  <piscisaureus_>creationix: so you could use uv_pipe routines to read from it
23:44:04  <piscisaureus_>instead of using the thread pool
23:44:18  <creationix>that would be nice
23:45:12  <piscisaureus_>https://github.com/joyent/libuv/blob/master/src/unix/pipe.c#L175
23:45:16  <creationix>so I create a pipe passing the fd from fs.open and treat it like a stream?
23:45:23  <piscisaureus_>yes
23:45:38  <piscisaureus_>In node there is no nice way to do it btw
23:45:38  <creationix>is there a way to tell it I want my reads to be 8 bytes at a time?
23:45:46  <piscisaureus_>creationix: hmm not really
23:46:21  <creationix>thanks
23:46:22  <piscisaureus_>creationix: you would have to make uv_alloc_cb return 8 byte sized buffers
23:47:29  <creationix>I see
23:50:13  * pieternquit (Quit: pietern)
23:51:33  <creationix>piscisaureus_, thanks. I'll stick to thread pool for now
23:51:34  <creationix>it works
23:54:17  <piscisaureus_>:-)
23:54:26  <piscisaureus_>creationix: winning argument
23:55:00  <isaacs>mmalecki: happy birthday!
23:55:22  <mmalecki>isaacs: thanks! :)
23:55:55  <piscisaureus_>crap. out of gas.
23:58:05  <isaacs>piscisaureus_: i thought you were signing off. go get some rest, man
23:59:52  * piscisaureus_quit (Read error: Operation timed out)