00:21:04  * bnoordhuisquit (Ping timeout: 240 seconds)
00:23:11  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
00:38:31  <ryah>piscisaureus___: people will need to ack msgs
00:38:53  <ryah>igorzi: heat? okay. :(
00:39:38  <ryah>piscisaureus___: is that the right link?
00:45:04  * piscisaureus_joined
00:49:10  * piscisaureus_quit (Client Quit)
00:50:55  * piscisaureus_joined
01:04:05  <ryah>piscisaureus___: no AMD
01:16:25  <piscisaureus_>ryah: ok - explain to me why
01:16:37  <piscisaureus_>ryah: btw i am not in favor of adding amd to core
01:17:08  <piscisaureus_>but when people start hacking node internals we may actually see how to help them
01:17:52  <piscisaureus_>ryah: or do you have reasons to block people from doing dat?
01:17:56  <piscisaureus_>*that?
01:31:27  * mikealjoined
01:40:51  * ericktquit (Quit: erickt)
02:00:45  * mikealquit (Quit: Leaving.)
02:14:57  * AvianFlu_joined
02:15:51  * AvianFluquit (Disconnected by services)
02:15:55  * AvianFlu_changed nick to AvianFlu
02:38:06  * dap1quit (Quit: Leaving.)
02:53:34  * brsonquit (Quit: leaving)
03:10:35  * AvianFluquit (Ping timeout: 260 seconds)
03:12:23  * ericktjoined
03:13:19  * AvianFlujoined
03:16:57  * ericktquit (Ping timeout: 244 seconds)
04:00:12  * ericktjoined
04:03:48  * isaacsquit (Quit: isaacs)
04:16:05  * dapjoined
04:20:34  * dapquit (Client Quit)
04:22:28  * dapjoined
04:49:25  * dapquit (Quit: Leaving.)
04:58:22  * dapjoined
05:09:36  * ericktquit (Quit: erickt)
05:14:23  * dapquit (Quit: Leaving.)
05:21:32  * wankdanker_quit (Remote host closed the connection)
05:30:37  <ryah>piscisaureus_: sorry - isaacs did a patch for that a few months ago and we ripped it out
05:47:01  * mikealjoined
05:54:55  * mikealquit (Ping timeout: 248 seconds)
05:55:15  * mikealjoined
06:11:15  * AvianFluquit (Ping timeout: 260 seconds)
06:31:15  * paddybyersquit (Quit: paddybyers)
06:32:00  * paddybyersjoined
06:38:36  * AvianFlujoined
06:39:28  * mikealquit (Ping timeout: 240 seconds)
06:51:39  * mikealjoined
06:53:23  * mikeal1joined
06:55:06  * mikeal1quit (Client Quit)
06:56:34  * mikealquit (Ping timeout: 252 seconds)
07:45:44  * AvianFluquit (Quit: Leaving)
08:28:50  * kuebkjoined
09:18:10  * paddybyers_joined
09:21:04  * paddybyersquit (Ping timeout: 240 seconds)
09:21:04  * paddybyers_changed nick to paddybyers
11:28:45  * piscisaureus_quit (Ping timeout: 260 seconds)
11:29:43  * bnoordhuisjoined
11:30:31  <bnoordhuis>piscisaureus___: https://github.com/joyent/node/blob/3f862cf/test/simple/test-fs-utimes.js#L101 <- why the different fs.open() flag on windows?
11:38:29  * piscisaureus_joined
11:40:50  <bnoordhuis>piscisaureus_: https://github.com/joyent/node/blob/3f862cf/test/simple/test-fs-utimes.js#L101 <- why the different fs.open() flag on windows?
11:42:19  <piscisaureus_>bnoordhuis: hmm I don't know
11:43:09  <piscisaureus_>maybe futimes doesn't work when it is open for reading only?
11:43:34  <bnoordhuis>piscisaureus_: i suspect it's something like that, i left a comment on igorzi's commit
11:43:47  <piscisaureus_>yeah... that seems the best thing to do
11:43:52  <piscisaureus_>it looks funky
12:12:39  <piscisaureus_>bnoordhuis: I think the windows crt for some reason makes a SetFileTime syscall after reading the time
12:12:59  <piscisaureus_>which would fail if we didn't open the file for writing
12:15:17  <piscisaureus_>oh wait... futime is for *setting* the time
12:15:19  <piscisaureus_>yeah
12:15:27  <piscisaureus_>that doesn't work if the file isn't open for writing
12:15:38  <bnoordhuis>i suppose that makes sense
12:16:11  <piscisaureus_>I wonder why this is not a requirement on unix
12:16:47  <bnoordhuis>because the file and its metadata are separated
12:17:14  <piscisaureus_>bnoordhuis: I think that is true on windows as well
12:17:14  <bnoordhuis>there's still access controls in effect, you just don't have to have the file opened in write mode
12:17:34  <piscisaureus_>bnoordhuis: I think we should specify the FILE_WRITE_ATTRIBUTES flag when opening a file for reading
12:17:59  <piscisaureus_>so we have GENERIC_READ | FILE_WRITE_ATTRIBUTES
12:18:40  <bnoordhuis>sounds good to me if it works
12:27:06  <piscisaureus_>bnoordhuis: I am not sure if that works with the way locking is handled in windows
12:28:01  <bnoordhuis>piscisaureus_: meaning you can read but not update attributes if another process has locked the file?
12:41:30  <piscisaureus_>bnoordhuis: it could mean that our open('r') call could fail if another process has locked the file for writing (so we can only read it).
12:42:28  <bnoordhuis>piscisaureus_: hmm, that's undesirable behaviour
12:56:43  * paddybyersquit (Remote host closed the connection)
12:58:35  * paddybyersjoined
13:04:45  * creationixjoined
14:04:08  <bnoordhuis>piscisaureus_: https://gist.github.com/1385726 <- sunos file watcher, something to play around with
14:04:27  <bnoordhuis>it's odd, with libev the event gets refired endlessly but not with plain event ports
14:05:06  <bnoordhuis>but plain event ports sucks too, it will report `echo 42 > foo` the first time (when it creates the file) but not the second time
14:05:28  <bnoordhuis>when you watch a directory, that is
14:15:26  <piscisaureus_>bnoordhuis: that is what we ran into as well
14:15:34  <piscisaureus_>bnoordhuis: is there no way to detect a file change?
14:16:32  <bnoordhuis>piscisaureus_: apparently not
14:19:02  <piscisaureus_>:-&
14:19:34  <creationix>what does spotlight on osx use?
14:19:42  <creationix>something apple specific?
14:21:59  <piscisaureus_>creationix: most likely kqueue and not event ports
14:22:07  <piscisaureus_>since event ports is a sunos thing
14:22:12  <bnoordhuis>or fsevents
14:22:23  <piscisaureus_>yup or that
14:23:44  <piscisaureus_>bnoordhuis: so FILE_MODIFIED effectively does nothing?
14:24:15  <bnoordhuis>piscisaureus_: only when a new file is created inside the watched directory
14:24:32  <bnoordhuis>but changes to an existing file aren't propagated
14:24:50  <piscisaureus_>hmm
14:24:54  <piscisaureus_>that sucks
14:25:07  <piscisaureus_>bnoordhuis: too bad nobody uses solaris
14:25:18  <piscisaureus_>so googling won't help
14:25:30  <bnoordhuis>hah
14:25:49  <bnoordhuis>solaris 10 doesn't have an event ports that supports fs watching
14:25:56  <bnoordhuis>that's an opensolaris / solaris 11 addition
14:26:05  <bnoordhuis>and solaris 11 got out only two weeks ago
14:28:09  <piscisaureus_>so maybe solaris 11 actually works right?
14:32:27  <bnoordhuis>piscisaureus_: i wouldn't bet a sixpack on it
14:37:01  <piscisaureus_>bnoordhuis: so what do we do with that now?
14:37:14  <piscisaureus_>bnoordhuis: throw when the user tries to use fs.watch on solaris?
14:37:40  <bnoordhuis>piscisaureus_: that's what happens right now, uv_fs_event_init() returns an error
14:37:41  <piscisaureus_>I mean, it's unreliable
14:37:48  <piscisaureus_>bnoordhuis: huh no
14:37:54  <piscisaureus_>bnoordhuis: it works on v0.6-latest
14:38:14  <bnoordhuis>you're running on illumos, not solaris
14:38:28  <piscisaureus_>ah ok
14:38:30  <bnoordhuis>solaris is what oracle ships, trademark and all
14:38:48  <piscisaureus_>bnoordhuis: what about you guys fix illumos? :-)
14:39:22  <bnoordhuis>bnoordhuis: i suppose that's an option :)
14:40:11  <bnoordhuis>but maybe we should simply use stat polling on anything that's not linux or windows
14:40:18  <bnoordhuis>(kqueue has issues too)
14:40:28  <piscisaureus_>bnoordhuis: windows has issues too as you know :-)
14:40:46  <piscisaureus_>bnoordhuis: but how would you monitor directories w/ stat polling?
14:41:07  <bnoordhuis>look if the directory's ctime changed, then scan it
14:41:15  <bnoordhuis>that's slow and race-y obviously
14:41:29  <piscisaureus_>bnoordhuis: does writing to a file update the directory's ctime?
14:41:34  <piscisaureus_>bnoordhuis: I suppose not
14:41:52  <bnoordhuis>hmm, i think it should but let me check
14:42:30  <bnoordhuis>it updates the mtime
14:42:40  <bnoordhuis>oh, and the ctime too
14:42:50  <bnoordhuis>let's double-check that on sunos just in case :)
14:42:53  <piscisaureus_>bnoordhuis: so that propagates all the way up to the root?
14:43:01  <piscisaureus_>bnoordhuis: that's actually pretty useful
14:43:24  <piscisaureus_>bnoordhuis: we could do recursive watching :-)
14:43:54  <bnoordhuis>piscisaureus_: no, it only updates the directory that contains the file
14:44:05  <bnoordhuis>it *will* update the atime all the way down to the root
14:44:13  <bnoordhuis>but file systems can be mounted with noatime
14:44:19  <piscisaureus_>yeah
14:44:35  <piscisaureus_>bnoordhuis: yeah it seems that on windows mtime gets updated too
14:44:57  <piscisaureus_>bnoordhuis: but ctime is not updated on windows
14:45:14  <piscisaureus_>dir a
14:45:29  <piscisaureus_>wrong box
14:45:45  <bnoordhuis>heh
14:46:28  <bnoordhuis>let me amend that remark about atime: it'll update the atime all the way down to the top-level directory of the mount
14:46:36  <bnoordhuis>i don't think it works across mount points
14:50:09  <bnoordhuis>err... maybe not, let's go with that atime is not a reliable measure
15:02:04  <CIA-95>node: Ben Noordhuis v0.6 * r7defbd2 / (4 files in 2 dirs): test: add DSA sign/verify crypto tests - http://git.io/O-HqPw
15:31:07  * CoverSlidequit (Read error: Connection reset by peer)
15:31:11  * CoverSlidejoined
15:38:44  * creationixquit (Remote host closed the connection)
16:10:49  <CIA-95>node: Ben Noordhuis isolates * rb20a257 / (16 files in 7 dirs): uv: upgrade to 0834e73 - http://git.io/uhrePw
16:10:50  <CIA-95>node: Ben Noordhuis isolates * r9f27c2c / (18 files in 2 dirs): isolates: isolate-ify the main loop - http://git.io/36uvbA
16:27:56  * isaacsjoined
16:29:32  * felixgejoined
16:29:32  * felixgequit (Changing host)
16:29:32  * felixgejoined
16:32:20  <felixge>did anybody have a chance to look over this yet? https://github.com/joyent/node/pull/2161
16:33:58  * kuebkpart
16:34:58  <bnoordhuis>felixge: yes
16:35:06  <bnoordhuis>it's an api change so it would have to land in master
16:35:11  <bnoordhuis>apart from that i defer to isaacs
16:35:14  <felixge>that was the plan
16:35:27  <piscisaureus_>felixge: the removals look good to me but I am not familiar enough with the code to know whether you forgot anything
16:35:29  <felixge>I think the decision to do that was already made: https://github.com/joyent/node/issues/1996
16:35:38  <piscisaureus_>but \o/ for seeing it go
16:35:56  <felixge>ok
16:36:01  <felixge>I'll wait for isaac to look as well
16:36:10  <isaacs>felixge: looks good, but can you rebase onto master?
16:36:34  <paddybyers>bnoordhuis: question re native add ons: what's going to be the official way of building these post-0.6?
16:36:47  <felixge>isaacs: it was rebased onto master when I pushed, wasn't it?
16:36:56  <piscisaureus_>hmm. terrible crashes with my new reqwrap :-(
16:37:25  <isaacs>felixge: hm... github seems to think it can't be automatically merged.
16:37:34  <felixge>isaacs: yeah, that's weird
16:37:44  <felixge>let me make sure
16:37:48  <isaacs>felixge: you rebased onto the latest joyent/master?
16:38:20  <bnoordhuis>paddybyers: undecided so far
16:38:40  <bnoordhuis>ryah wants to push for binary distribution of modules
16:39:03  <piscisaureus_>... which does not make the problem go away
16:39:07  <bnoordhuis>but i don't want that for unices
16:39:33  <felixge>isaacs: it's rebased properly
16:39:40  <isaacs>felixge: weird.
16:39:40  <felixge>I think it happened because I did a force-push to fix a typo
16:39:42  <felixge>in my branch
16:39:45  <isaacs>oic
16:39:48  <isaacs>hm.
16:39:50  <paddybyers>bnoordhuis: I'm doing an example with a gyp build file, but having a few issues (e.g. persuading Xcode via gyp to give it the correct name/extension)
16:40:01  <felixge>isaacs: want me to merge add the commit to master?
16:40:14  <isaacs>lemme see if it merges cleanly for me
16:40:23  <felixge>k
16:40:24  <isaacs>(and test, though this is somewhat trivial)
16:40:30  <felixge>isaacs: thanks a bunch
16:40:36  <bnoordhuis>paddybyers: for 0.6 we're sticking to node-waf
16:41:05  <bnoordhuis>i don't really have any well-formed ideas on how to do it in 0.8 with gyp
16:41:46  <bnoordhuis>(read: open to suggestions)
16:41:52  <isaacs>make test running
16:42:37  <paddybyers>bnoordhuis: (maybe I'm being dumb) but I think I have to have node as a shared library and link the add-on against that; I can't link it against the node executable
16:43:30  <paddybyers>bnoordhuis: it sounds like something similar was done already for win32
16:43:48  <bnoordhuis>paddybyers: maybe... compiling (part of) node to a shared object creates a new set of issues though
16:44:16  <bnoordhuis>on unices shared objects are only loaded from a predefined set of directories + LD_LIBRARY_PATH
16:44:26  <bnoordhuis>well, i suppose it's one option
16:45:23  <paddybyers>bnoordhuis: on darwin it's different, and worse, because by default the shared lib has its absolute location hard-wired into the binary itself
16:46:09  <paddybyers>I'll keep playing around; provided I'm not missing something obvious
16:46:43  <bnoordhuis>paddybyers: oh right, it works the same on e.g. linux - i should've said 'linked against' instead of 'loaded from'
16:47:16  <bnoordhuis>or... well, you still need to set LD_LIBRARY_PATH if you've linked against a .so that's outside the regular directories
16:47:46  <bnoordhuis>once again no silver bullet
16:49:06  <isaacs>wow, throwing on process.stdout.end() is really harsh.
16:49:06  <isaacs>s
16:49:19  <isaacs>that means you can't pipe to stdout.
16:49:26  <paddybyers>bnoordhuis: I found this which looks like it could work; you can put relative paths into the shared library and the executable that depends on it: http://blogs.oracle.com/dipol/entry/dynamic_libraries_rpath_and_mac
16:49:45  <isaacs>making it a no-op is one thing, but throwing is too aggressive, i think
16:50:21  <isaacs>felixge: your patch looks good, does something we've all discussed and want, and doesn't cause any new test breakages. i'm landing it.
16:50:45  <felixge>isaacs: cool
16:50:59  <felixge>next on my list: try to see if I can get a few osx fails fixed
16:51:05  <felixge>I get a lot of failing tests on my machine
16:51:13  <felixge>which is rather disturbing
16:51:17  <felixge>(a lot = ~6-7 or so)
16:51:31  <bnoordhuis>paddybyers: hah, that's so evil - bloody macs
17:02:40  * dapjoined
17:02:42  <CIA-95>node: isaacs master * re543b0e / test/simple/test-zlib-from-string.js : zlib: Fix invalidly failing test - http://git.io/Rc5MjA
17:02:47  <CIA-95>node: Felix Geisendörfer master * rb3f91f1 / (lib/stream.js test/simple/test-stream-pipe-cleanup.js):
17:02:47  <CIA-95>node: Remove support for multi-source pipe()
17:02:47  <CIA-95>node: This reverts 6c5b31bd which had too few use cases, too much complexity,
17:02:47  <CIA-95>node: and can be handled in user-land by using `{end: false}`.
17:02:47  <CIA-95>node: Closes #1996 - http://git.io/Pp7icQ
17:12:26  * ericktjoined
17:20:30  <ryah>yo
17:20:39  <ryah>bnoordhuis: you added rsa support to crypto a while back?
17:21:03  <bnoordhuis>ryah: signing/verification yes
17:22:16  <bnoordhuis>dinner, biab
17:23:55  <ryah>we should do some perf testing on https at some point. does 3x slower throughput sound reasonable?
17:24:19  * ryahneeds to setup nginx and test
17:25:08  <ryah>https://github.com/joyent/node/issues/2061 <-- damn!
17:29:58  * ericktquit (Quit: erickt)
17:31:52  <ryah>igorzi: https://twitter.com/#!/kevm/status/138935683355119616
17:32:05  <ryah>igorzi: did your friend mention Paraffin?
17:37:43  <ryah>piscisaureus_: yt?
17:37:51  <piscisaureus_>ryah: yup
17:38:01  <piscisaureus_>ryah: the new Wrap stuff still crashes
17:38:01  <ryah>piscisaureus_: where are the WiX files? :)
17:38:12  <ryah>i'm looking for this heat.exe...
17:38:13  <piscisaureus_>ryah: eh, what?
17:38:37  <piscisaureus_>At sourceforge I think
17:38:51  <ryah>piscisaureus_: i have it installed - but i dont know where they are on my computer..
17:39:03  <ryah>i want 'dpkg -L' if you know what i mean
17:39:08  <piscisaureus_>aah
17:39:09  <piscisaureus_>hmm
17:39:34  <piscisaureus_>let me look it up
17:40:08  <ryah>nevermind i found it
17:40:10  <piscisaureus_>ryah: you can also search if you have some spare time
17:40:21  <piscisaureus_>ryah: `dir /s/b c:\heat.exe`
17:41:32  <ryah>nice thanks
17:42:38  <ryah>C:\Program Files\Windows Installer XML v3.5\bin
17:42:57  <ryah>windows really needs make
17:43:09  <ryah>this vcbuild.bat shit is not sufficent
17:43:26  <piscisaureus_>what's missing?
17:43:58  <ryah>the ability to check if something's modified
17:44:08  <piscisaureus_>eh
17:44:13  <piscisaureus_>why? it does that
17:48:00  <ryah>actually it wouldnt help me in this case
18:02:21  * ericktjoined
18:04:10  * sh1mmerjoined
18:05:57  * piscisaureus_part
18:14:30  * piscisaureus_joined
18:14:38  * piscisaureus_changed nick to piscisaureus
18:24:26  <bnoordhuis>ryah: only 3x slower would make me a happy man, it's often more like 10x
18:26:38  <CIA-95>node: Ryan Dahl v0.6 * r6035651 / doc/api/tty.markdown : Fixes #2160. Fix tty.getWindowSize docs - http://git.io/FCvU-g
18:26:59  <ryah>piscisaureus, igorzi: is nodemsi.sln something that can be opened in visual studio?
18:27:06  <ryah>i can't seem to open it...
18:29:44  * brsonjoined
18:32:11  <igorzi>ryah: yeah, you should be albe to open it in vs.. it won't really be useful though, it'll just open the .wxs file in the xml editor
18:33:22  <igorzi>ryah: Paraffin.. no, my friend didn't mention it. have you tried it?
18:34:06  <ryah>igorzi: no, im using heat - seems reasonable
18:42:38  * AvianFlujoined
18:47:47  <paddybyers>bnoordhuis: add-on loading doesn't follow symlinks properly - is this a known issue?
18:48:04  <paddybyers>e.g. if hello.node -> libhello.dylib
18:48:32  <paddybyers>then the change of extension confuses it
18:48:37  <bnoordhuis>paddybyers: that's probably a feature of your system's dlopen() function
18:49:21  <paddybyers>no, it's because once it's found the file, it uses the extension of the source file instead of the target to decide what to do; it doesn't even attempt to dlopen
18:49:50  <paddybyers>if the file is just copied, it works
18:50:10  <paddybyers>if it's intentional then fine, otherwise I'll have a look and see what's happening
18:52:07  * piscisaureusquit (Quit: ~ Trillian Astra - www.trillian.im ~)
18:52:32  <ryah>paddybyers: it's not intentional
18:52:43  <ryah>i suppose that could be fixed..
18:52:59  <paddybyers>ryah: ok, I'll have a look
18:53:05  <ryah>paddybyers: cool
18:59:14  * piscisaureus_joined
19:00:14  <igorzi>piscisaureus_: what do you think about this: https://gist.github.com/1386547 ?
19:01:12  * mikealjoined
19:18:39  <ryah>piscisaureus_: oops i quit skype
19:18:42  <ryah>piscisaureus_: let's chat now
19:18:50  <piscisaureus_>ryah: I have to reboot first
19:18:54  <piscisaureus_>it'll take 5 minutes
19:18:56  <ryah>ok
19:19:06  <piscisaureus_>because my scrfeen is borked after a driver upgrade that went wrong
19:19:18  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
19:19:33  <ryah>im getting a coffee bbiab
19:20:30  * piscisaureus_joined
19:41:52  <isaacs>hrm... i'm really strongly considering just monkey-patching fs.open and fs.close to get around this ulimit stuff.
19:43:25  * ericktquit (Quit: erickt)
19:54:38  * AvianFlu_joined
19:55:00  * AvianFluquit (Disconnected by services)
19:55:04  * AvianFlu_changed nick to AvianFlu
20:01:14  <bnoordhuis>signing off early today, need to catch up on some sleep
20:01:17  <bnoordhuis>see you guys tomorrow
20:01:43  <ryah>bnoordhuis: night!
20:01:58  <ryah>bnoordhuis: is the operator* overloading really necessary?
20:03:06  <piscisaureus_>igorzi: the patch kinda looks good...
20:03:27  <piscisaureus_>igorzi: but are you sure it still accepts forward slashes ?
20:05:25  * bnoordhuisquit (Ping timeout: 240 seconds)
20:05:30  <piscisaureus_>ryah: I actually lost the linked list code because I incorrectly assumed that git stash would have saved it
20:05:55  <piscisaureus_>which is probably fine, I will do it again now.
20:07:19  <ryah>piscisaureus_: ok
20:07:33  <piscisaureus_>nah. not ok.
20:07:56  <piscisaureus_>let me see how far my undo stack goes back, maybe I can get it back anyway
20:12:07  <felixge>is test/simple/test-http-response-no-headers.js failing for anybody else?
20:12:11  <felixge>I get a http parse error
20:13:46  <ryah>felixge: i think that's a known failure
20:14:04  <felixge>ryah: I can look into fixing it
20:14:14  <felixge>seems like the http parser doesn't ilke http 0.9 for one
20:14:19  <felixge>* like
20:14:28  <felixge>should that be valid?
20:15:32  <felixge>i guess it should
20:15:34  <felixge>will play with this a little
20:16:02  <ryah>felixge: see https://github.com/joyent/node/commit/4f38c5ece3017923eebf034b815755c98c8c6113
20:16:20  <ryah>and https://github.com/joyent/node/issues/1711
20:17:32  <felixge>looking into it
20:17:34  <felixge>thanks for the pointers
20:20:36  <ryah>felixge: cool
20:23:05  * DrPizzaquit (Ping timeout: 260 seconds)
20:28:42  * DrPizzajoined
20:28:45  <felixge>ryah: https://github.com/felixge/http-parser/commit/4889d391b0047339bc684fbfea1af5bd6b3bbb30
20:28:49  <felixge>^-- fixes the 0.9 test
20:29:04  <felixge>1.1 still fails, but I'm looking into that now
20:29:29  <felixge>oh, forgot to change the test name
20:29:30  <felixge>will fix that
20:30:28  <felixge>https://github.com/joyent/http-parser/commit/1f58837e39f3fc64bf16785dd45e6c2c7146b1fb
20:30:30  <felixge>^- better
20:32:27  <felixge>is http 1.1 supposed to accept no headers + body?
20:32:52  <felixge>ryah: from what you said in the comment it's supposed to assume keep-alive which means we must assume not body if we can't determine the length of it, right?
20:33:00  <felixge>* assume no body
20:38:28  <ryah>felixge: one sec
20:40:50  <ryah>felixge: right
20:41:02  <ryah>felixge: 1.1 assumes keep-alive
20:41:22  <CIA-95>node: Ryan Dahl v0.6 * r2cd5adc / (tools/wafadmin/Build.py wscript): Install npm with 'make install' - http://git.io/IjUNlw
20:41:31  * piscisaureus_quit (Ping timeout: 260 seconds)
20:43:09  <felixge>ryah: ok, I'll modify the test
20:44:37  <CIA-95>node: Ryan Dahl v0.6 * r38c49fb / (tools/wafadmin/Build.py wscript): Fix broken commit 2cd5adc - http://git.io/jKPD3A
20:45:02  <felixge>ryah: how do I upgrade the http parser, just copy the .c & .h ?
20:45:06  <felixge>or everything?
20:45:11  <felixge>where does the gyp come from?
20:45:32  <ryah>felixge: usually i use git-archive
20:45:50  <ryah>felixge: also sorry - but you need to sign a separate CLA for http-parser https://github.com/joyent/http-parser/blob/master/CONTRIBUTIONS
20:45:56  <ryah>which is actually the same
20:46:06  <ryah>but for some reason they are managing it differently
20:47:00  <felixge>ryah: done
20:48:16  <ryah>git archive --prefix=http_parser/ HEAD | (cd ~/projects/node/deps; tar -xf - )
20:50:33  <felixge>ryah: https://github.com/joyent/http-parser/commit/1f58837e39f3fc64bf16785dd45e6c2c7146b1fb
20:50:37  <felixge>do you want to pull this in first?
20:50:46  <felixge>so I can reference the sha1 ref from the joyent repo?
20:51:08  <felixge>https://github.com/joyent/http-parser/commit/1f58837e39f3fc64bf16785dd45e6c2c7146b1fb.patch
20:51:36  <ryah>felixge: okay updated http-parser master
20:52:14  <ryah>felixge: will you update http-parser in node-v0.6 ?
20:52:30  <felixge>I was going to put this in master
20:52:39  <felixge>but I guess it's up to you where you merge this in
20:52:54  <ryah>*shrug* i think it's okay to put in v0.6 if the diff isn't great
20:52:58  <ryah>let me see...
20:54:43  <ryah>felixge: i'll update it
20:55:11  <isaacs>d'oh. lchmod doesn't close the file descriptor.
20:56:14  * isaacsfires up emacs to write some hate mail to whoever wrote 3935adce...
20:56:35  <CIA-95>node: Ryan Dahl v0.6 * r3abebfe / (6 files): Upgrade http-parser to joyent/[email protected] - http://git.io/gLCpHg
20:57:40  <ryah>now i'll merge v0.6 into master
20:57:43  <felixge>ryah: https://github.com/joyent/node/pull/2165
20:57:44  <ryah>so many fucking steps
20:57:51  <felixge>running tests again now
20:57:53  <felixge>don't merge yet
20:57:55  <felixge>but have a look at it
20:58:35  <ryah>felixge: im just going to merge v0.6 - you can land that test fix yourself after
20:58:58  <felixge>in 0.6 or master?
21:00:38  * piscisaureus_joined
21:01:27  <ryah>felixge: 0.6 into master
21:01:37  <ryah>felixge: that test is only in master - so land it there
21:01:45  <ryah>felixge: we might want to ahve bnoordhuis review?
21:01:52  <ryah>nah, it looks okay
21:02:08  <ryah>im just building master now to test.. which will take a while
21:02:40  * mikealquit (Quit: Leaving.)
21:02:44  <felixge>ok, my tests finished in master, looks good (down to 7 failing tests on osx now)
21:02:59  <felixge>ping me when I can merge
21:03:01  <felixge>headed out for a sec
21:04:11  <felixge>oh and tell me if you want me to squash or not
21:04:12  <felixge>brb
21:12:30  <isaacs>can anyone review this?
21:12:47  <isaacs>i'm not sure the best way to test it without hijacking the fs.open and fs.close methods
21:12:58  * igorziquit (Ping timeout: 265 seconds)
21:13:08  <isaacs>https://github.com/isaacs/node/commit/d18087235fd18d41b4059eef033a394572a291c6
21:16:00  <felixge>isaacs: I see nothing wrong with hijacking open/close
21:16:29  <felixge>patch LGTM
21:16:32  <felixge>one issue so
21:16:35  * igorzijoined
21:16:35  <felixge>I'd do err || err2
21:16:52  <felixge>also, you should wrap fchmod in try...catch
21:16:53  <igorzi>piscisaureus_: i'll test with forward slashes
21:17:00  <felixge>to enable the same logic there
21:17:13  <felixge>(as in: still try to close the fd, even if chmod failed)
21:17:43  <isaacs>felixge: good call.
21:19:00  * mikealjoined
21:25:55  <ryah>god damn it. im getting a strange segfault...
21:26:57  <ryah>it MIGHT be related to https://github.com/joyent/node/issues/2061
21:29:27  <ryah>it might be a V8 problem
21:30:31  <CIA-95>node: Ryan Dahl master * r6035651 / doc/api/tty.markdown : Fixes #2160. Fix tty.getWindowSize docs - http://git.io/FCvU-g
21:30:32  <CIA-95>node: Ryan Dahl master * r2cd5adc / (tools/wafadmin/Build.py wscript): Install npm with 'make install' - http://git.io/IjUNlw
21:30:32  <CIA-95>node: Ryan Dahl master * r38c49fb / (tools/wafadmin/Build.py wscript): Fix broken commit 2cd5adc - http://git.io/jKPD3A
21:30:32  <CIA-95>node: Ryan Dahl master * r3abebfe / (6 files): Upgrade http-parser to joyent/[email protected] - http://git.io/gLCpHg
21:30:34  <CIA-95>node: Ryan Dahl master * r8595981 / (483 files in 84 dirs):
21:30:34  <CIA-95>node: Merge remote branch 'origin/v0.6'
21:30:34  <CIA-95>node: Conflicts:
21:30:34  <CIA-95>node: Makefile
21:30:35  <CIA-95>node: src/node_version.h
21:30:35  <CIA-95>node: test/simple/test-crypto.js - http://git.io/qEgolw
21:35:30  <isaacs>review? https://github.com/isaacs/node/commit/d4ad93dc52c8781bb8448a71a829af3e472b2ced
21:35:37  <isaacs>felixge: made the changes you suggested.
21:35:40  <isaacs>test passes.
21:38:09  <felixge>isaacs: LGTM
21:42:54  <ryah>isaacs: land that in v0.6
21:43:05  <ryah>felixge: you can land your test fix now
21:43:26  <felixge>ryah: ok. 0.6 or master? squashed, or unsquashed?
21:43:52  <ryah>felixge: the test only exists in master. however makes sense
21:44:16  <felixge>I can cherry-pick the test into 0.6
21:45:24  <ryah>if you want, sure
21:45:33  <felixge>ok
21:45:34  <felixge>will do
21:51:09  <CIA-95>node: Ben Noordhuis v0.6 * r246805d / test/simple/test-http-response-no-headers.js :
21:51:09  <CIA-95>node: test: add 'response body with no headers' http test
21:51:09  <CIA-95>node: HTTP/0.9 - fails with a parse error
21:51:09  <CIA-95>node: HTTP/1.0 - works
21:51:09  <CIA-95>node: HTTP/1.1 - fails with an empty response body
21:51:10  <CIA-95>node: See #1711. - http://git.io/kb2B2A
21:51:10  <CIA-95>node: Felix Geisendörfer v0.6 * r6e1e9e2 / test/simple/test-http-response-no-headers.js :
21:51:11  <CIA-95>node: Get test-http-response-no-headers.js to pass
21:51:12  <CIA-95>node: Main fix was in 3abebf which added HTTP/0.9 support to http parser.
21:51:12  <CIA-95>node: Changed test because HTTP 1.1 mandates keep-alive when no headers are
21:51:13  <CIA-95>node: given.
21:51:13  <CIA-95>node: Fixes #1711 - http://git.io/EsacLA
21:51:45  <felixge>^--
21:51:49  <felixge>done
21:53:59  <ryah>k
22:00:10  <felixge>question: there are a ton of pull requests open, would you like me to go over these and close the unreasonable ones / recommend some we may want to land?
22:00:59  <ryah>felixge: that would be great
22:01:09  <CIA-95>node: isaacs v0.6 * r0ba78d5 / (lib/fs.js test/simple/test-fs-chmod.js): Close #2166 Close the fd in lchmod - http://git.io/vaCbiA
22:01:22  <felixge>ryah: ok, will start with this tomorrow
22:01:31  <paddybyers>ryah: first attempt at fixing that issue with symlinks that change extension: https://github.com/paddybyers/node/commit/119b2c079f2a3c6078591bd3b0568848f16b0eba
22:03:58  <ryah>paddybyers: okay - you should have isaacs review this
22:04:05  <ryah>paddybyers: is it possible to have a test?
22:04:29  * isaacsreviewing...
22:05:14  <piscisaureus_>http://polarssl.org/trac/wiki/HighLevelDesign#scenario_b
22:05:24  <piscisaureus_>^-- pquerna: looks pretty neat
22:05:40  <piscisaureus_>they actually made an attempt to make it readable
22:05:55  <paddybyers>ryah: yes, but not tonight :)
22:06:36  <piscisaureus_>http://polarssl.org/source_code
22:06:51  <isaacs>paddybyers: it seems like this could be done a bit more simply...
22:07:30  <isaacs>paddybyers: the goal is that if you, eg symlink: `ln -s foo.js foo.json`, then require("./foo.json") should load with the .js ext instead of the .json extension, yes?
22:07:47  <paddybyers>isaacs: yes
22:08:13  <paddybyers>isaacs; actually no, the other way around
22:08:57  <paddybyers>my scenario was ln -s lib hello.dylib hello.node
22:09:07  <paddybyers>needs to be handled as .node
22:09:57  <ryah>http://habrahabr.ru/blogs/nodejs/132745/ <-- checkout these numbers
22:10:42  <isaacs>paddybyers: i see....
22:10:58  <ryah>polarssl - awesome
22:11:14  <isaacs>paddybyers: why can't we just make dylib load with dlopen as well?
22:11:14  <ryah>fucking GPL though
22:12:30  <paddybyers>isaacs; not sure what you mean .. you mean just add Module._extensions['.dylib'] = … ?
22:12:38  <isaacs>h
22:12:41  <isaacs>paddybyers: yeah
22:12:57  <isaacs>paddybyers: Module._extensions['.dylib'] = Module._extensions['.node']
22:13:40  <piscisaureus_>ryah: they have an exception program - but I dont really understand it
22:13:54  <piscisaureus_>it seems they allow it to be used with BSD licences too
22:14:22  <paddybyers>isaacs: that would work
22:14:40  * benviequit (Ping timeout: 240 seconds)
22:14:42  <igorzi>ryah: and that's running with iisnode
22:16:17  <paddybyers>isaacs: it depends I guess whether it's potentially an issue with other module types as well, or just with shared libraries
22:16:47  <isaacs>paddybyers: yeah... if you ln -s foo.coffee foo.js, I'd rather it be opened as coffeescript rather than javascript
22:17:23  <piscisaureus_>My russian is way too rusty :-(
22:17:25  <isaacs>or json, etc.
22:17:48  <paddybyers>isaacs: btw I couldn't help noticing that this function: https://github.com/joyent/node/blob/master/lib/module.js#L319
22:18:26  <paddybyers>always returns identical values in result[0] and result[1]; I guess there was some functionality there before that's disappeared
22:18:40  <isaacs>hahah
22:18:41  <isaacs>indeed.
22:19:09  <isaacs>feel free to send a pull req cleaning that up
22:19:46  <paddybyers>isaacs: so, your preference is in fact for the current behaviour for .js etc, but a special case for perhaps dylib, dll, etc
22:21:05  <isaacs>paddybyers: yeah
22:21:20  <isaacs>basically, just add those to the list of what we support
22:21:30  <paddybyers>isaacs: ok, np, will look at it again tomorrow, off now
22:21:35  <isaacs>k, sleep well
22:23:26  <ryah>http://babelfish.yahoo.com/translate_url?doit=done&tt=url&intl=1&fr=bf-home&trurl=http%3A%2F%2Fhabrahabr.ru%2Fblogs%2Fnodejs%2F132745%2F&lp=ru_en&btnTrUrl=Translate <-- translation
22:27:46  <igorzi>ryah: the first benchmark graph says "output of current time" (just translating the image)
22:28:12  <piscisaureus_>http://polarssl.org/license_exception <-- even WTFPL is fine for them
22:29:31  <piscisaureus_>"I hope that in the future the support of fibers (fibers) will be build it Node.js directly."
22:32:30  <ryah>piscisaureus_: i think that's a mistranslation
22:32:55  <ryah>piscisaureus_: it says "i'm am glad the core team will not support fibers in the future"
22:33:25  <piscisaureus_>ryah: wishful thinking :-)
22:34:41  <ryah>okay.. now what was i doing.. :/
22:34:43  <ryah>oh yeah msi
22:36:02  * benviejoined
22:41:18  <isaacs>ryah: what version is going to have that lchmod fix?
22:41:25  <isaacs>0.6.3?
22:49:01  * mikealquit (Quit: Leaving.)
22:56:03  <ryah>isaacs: did you land it in v0.6? then yest
22:56:06  <ryah>s/yest/yes
22:57:48  * creationixjoined
22:58:27  * creationixloves the topic
22:58:44  <piscisaureus_>http://www.twitter.com/tanepiper/status/138741778709876736
23:00:18  * piscisaureus_summons the http experts
23:00:59  <isaacs>ryah: i did
23:05:26  * AvianFluquit (Ping timeout: 255 seconds)
23:08:01  * paddybyersquit (Quit: paddybyers)
23:08:44  * paddybyersjoined
23:10:49  <ryah>isaacs: https://twitter.com/#!/tanepiper/status/138741778709876736 <--
23:12:56  <piscisaureus_>ryah: http://www.twitter.com/izs/status/139116499641171968
23:13:04  <piscisaureus_>metaconversation
23:15:56  <felixge> sleep, gn8
23:15:57  * felixgequit (Quit: felixge)
23:16:10  <piscisaureus_>night felixge
23:16:29  <isaacs>ryah: just discussed in #node.js
23:16:32  <isaacs>user error.
23:16:49  * felixgejoined
23:17:02  <isaacs>ryah: https://github.com/tanepiper/node-bitly/blob/master/bitly.js#L55
23:17:06  <isaacs>that's the 0.4 fix he added
23:19:19  * felixgequit (Client Quit)
23:20:23  * AvianFlujoined
23:29:30  * creationixquit (Read error: Operation timed out)
23:42:43  <ryah>wix is so awful
23:42:47  <ryah>what's wrong with these people
23:43:26  <ryah>it's hard to imagine that given the task of unpacking some files, they created the most convoluted solution possible
23:43:49  <ryah>*hard to not imagine
23:45:24  * bnoordhuisjoined
23:48:19  <piscisaureus_>maybe you can use the stuff that's built into msvs
23:48:32  <piscisaureus_>I don't know why wix was chosen
23:48:40  <piscisaureus_>maybe msvs is even more awful
23:57:43  * paddybyersquit (Quit: paddybyers)