00:00:00  <CIA-95>libuv: Bert Belder v0.6 * r45b976a / src/win/error.c : Win: add mapping for ERROR_ACCESS_DENIED - http://git.io/SA1VSQ
00:02:28  <CIA-95>node: Bert Belder v0.6 * rd51a0c3 / deps/uv/src/win/error.c : Upgrade libuv to 45b976a - http://git.io/MIJr0Q
00:03:28  * igorzijoined
00:04:20  <igorzi>piscisaureus: looks like the long-paths patch breaks require('./foo'), with foo/index.js
00:05:31  <igorzi>piscisaureus: i'll try to track it down before the release (i'll be on and off line), but if not it'll have to wait till 0.6.4
00:08:05  <piscisaureus>igorzi: nice work
00:08:12  <piscisaureus>igorzi: I will see it tomorrow
00:09:04  <igorzi>ahh actually, i see why it's breaking fs.stat('somedir').isDirectory() is returning false
00:09:16  <piscisaureus>ah ok
00:10:13  <piscisaureus>bnoordhuis: does test-fs-fsync pass on linux?
00:10:53  <bnoordhuis>piscisaureus: yep
00:11:03  <piscisaureus>bnoordhuis: weird
00:11:16  <piscisaureus>bnoordhuis: because test-fs-chmod changes fixtures/a.js to read-only
00:11:35  <piscisaureus>bnoordhuis: and then test-fs-fsync tries to open it in append mode
00:13:37  <bnoordhuis>piscisaureus: hmm, i see your point - but it does work here
00:15:14  <bnoordhuis>piscisaureus: https://github.com/joyent/node/blob/d51a0c3/test/simple/test-fs-chmod.js#L67 <- that's probably why
00:15:39  <piscisaureus>ah yeah
00:16:11  <piscisaureus>Let me change the order of rw and ro modes :-)
00:18:04  <piscisaureus>bnoordhuis: I also do not agree with this test
00:18:15  <piscisaureus>it chmods and fchmods the same file simultaneously
00:19:12  <bnoordhuis>so it does
00:19:44  <piscisaureus>bnoordhuis: or do you think it is good to test this?
00:19:54  <bnoordhuis>piscisaureus: no, it's wrong
00:19:57  <piscisaureus>bnoordhuis: it kind of demonstrates that race conditions do exist in node
00:20:44  <bnoordhuis>piscisaureus: libeio's task queue probably mitigates it somewhat but yeah
00:20:56  <piscisaureus>there are multiple threads
00:21:01  <piscisaureus>that's the pain
00:21:13  <piscisaureus>let me fix that test
00:26:13  <CIA-95>node: Bert Belder reviewme * r236b217 / test/simple/test-fs-chmod.js :
00:26:13  <CIA-95>node: Fix issues with test-fs-chmod
00:26:13  <CIA-95>node: - The test simultaneously chmods and fchmods the same file.
00:26:13  <CIA-95>node: - On windows, it leaves behind a fixture in read-only mode,
00:26:13  <CIA-95>node: which causes test-fs-fsync to fail. - http://git.io/Gas9jQ
00:26:22  <piscisaureus>bnoordhuis: --^ review
00:26:24  <piscisaureus>?
00:28:19  <bnoordhuis>piscisaureus: lgtm
00:28:56  <CIA-95>node: Bert Belder v0.6 * r236b217 / test/simple/test-fs-chmod.js :
00:28:57  <CIA-95>node: Fix issues with test-fs-chmod
00:28:57  <CIA-95>node: - The test simultaneously chmods and fchmods the same file.
00:28:57  <CIA-95>node: - On windows, it leaves behind a fixture in read-only mode,
00:28:57  <CIA-95>node: which causes test-fs-fsync to fail. - http://git.io/Gas9jQ
00:56:01  <bnoordhuis>piscisaureus: good luck with the release today
00:56:11  <bnoordhuis>i'm off to bed
01:00:36  * bnoordhuisquit (Ping timeout: 252 seconds)
01:00:56  <igorzi>piscisaureus: https://gist.github.com/1392592
01:01:11  <igorzi>we need to get off crt
01:05:50  <piscisaureus>igorzi: lgtm
01:06:07  <CIA-95>libuv: Bert Belder reviewme * r3a50f8f / src/win/getaddrinfo.c : Win: fix uv_getaddrinfo error reporting - http://git.io/150iLQ
01:06:20  <piscisaureus>igorzi: ^-- review that?
01:08:31  <igorzi>piscisaureus: lgtm
01:08:43  <piscisaureus>igorzi: yes we need to get off crt
01:09:29  <CIA-95>libuv: Bert Belder v0.6 * r3a50f8f / src/win/getaddrinfo.c : Win: fix uv_getaddrinfo error reporting - http://git.io/150iLQ
01:12:37  <piscisaureus>igorzi: we also need to implement realpath/lstat I think.
01:13:04  <piscisaureus>igorzi: but... later. first enjoy thanksgiving :-)
01:14:04  <CIA-95>node: Bert Belder v0.6 * r95eb729 / deps/uv/src/win/getaddrinfo.c : Upgrade libuv to 3a50f8f - http://git.io/rx4C4g
01:15:54  <piscisaureus>igorzi: but I honestly, the windows crt really looks like it was done in a hurry, and nobody cared to fix it afterwards
01:20:59  * brsonquit (Quit: leaving)
01:25:34  * piscisaureusquit (Quit: ~ Trillian Astra - www.trillian.im ~)
01:34:54  * indexzerojoined
01:55:04  * indexzeroquit (Ping timeout: 244 seconds)
02:10:52  <CIA-95>node: Bert Belder v0.6 * r232e8e1 / test/fixtures/echo.js : Fix test-stdin-from-file - http://git.io/TUAvJQ
02:20:04  * piscisaureus__joined
02:22:19  <piscisaureus__>Lion: 3 tests failing. 2 fail due to broken apache bench.
02:22:27  <piscisaureus__>proctitle fails too
02:38:21  <igorzi>piscisaureus__: now when fs functions fail, the err.path contains '\\?\' prefix. we should probably fix that? what do you think?
02:50:58  <igorzi>piscisaureus_: --^
03:21:41  <CIA-95>libuv: Igor Zinkovsky v0.6 * rcfa4112 / src/win/fs.c : windows: make uv_fs_stat use open + fstat to support long paths - http://git.io/Jvx0jw
03:28:16  <igorzi>piscisaureus_: https://gist.github.com/1392759
03:28:29  <igorzi>piscisaureus_: new patch..see node_file.cc
03:36:23  * indexzerojoined
03:40:55  <igorzi>piscisaureus_: i'm going offline... please land the https://gist.github.com/1392759 patch if it looks ok to you. i ran test-all with it, and there are no new failures caused by the patch
03:46:01  <CIA-95>node: Igor Zinkovsky v0.6 * r3195809 / deps/uv/src/win/fs.c : update libuv to cfa4112950b1f897fda7 - http://git.io/JaCANw
03:46:19  * igorziquit (Quit: Page closed)
04:47:39  * ericktjoined
06:31:40  * ericktquit (Quit: erickt)
06:32:20  * paddybyersjoined
06:53:55  * eddybjoined
06:54:08  <eddyb>random question:
06:54:57  <eddyb>is there a wrapper around libuv, for C++?
06:56:25  <eddyb>(I understand it was written for node, and there's no need for fancy C++ there)
06:56:46  <eddyb>or will libuv remain a platform layer for node?
07:14:22  * indexzeroquit (Ping timeout: 244 seconds)
07:25:36  * felixgejoined
07:25:36  * felixgequit (Changing host)
07:25:36  * felixgejoined
07:36:54  * paddybyersquit (Quit: paddybyers)
07:41:06  <ryah>piscisaureus__: yo
07:42:06  <ryah>eddyb: no
07:42:17  <ryah>eddyb: no wrapper for c++ that is
07:43:58  <eddyb>so it's just a posix/unix/linux-like platform layer?
07:44:12  <ryah>eddyb: ?
07:44:36  <eddyb>like, makes things easier for node on platforms that do not have that interface
07:44:42  <eddyb>also provides async everywhere
07:45:11  <eddyb>I'm sure things would be better if at least some of the libuv code was C++
07:45:15  <eddyb>but it's not my call
07:45:27  <ryah>?
07:45:37  <ryah>it's a platform layer, yes
07:45:38  <ryah>:)
07:45:48  <ryah>no, it would not be better if some of it was c++
07:46:01  <eddyb>wh?
07:46:07  <eddyb>*why
07:46:22  <ryah>because there's no need for it
07:47:48  <eddyb>c'mon...
07:47:55  <eddyb>even namespaces
07:48:13  <piscisaureus__>ryah: yo. I managed to get up
07:48:22  <piscisaureus__>ryah: let me get to the office, be there in 10 minutes
07:49:12  <ryah>ok
07:49:25  <ryah>im running tets
07:51:38  <piscisaureus__>ryah: 2 issues remain on my side:
07:51:38  <piscisaureus__>- it seems that unix now never reports EAINONAME (does ENOTFOUND instead). This needs to be fixed in windows.
07:51:38  <piscisaureus__>- I need to land the long-files patch by igor and create a small handler that strips a prefix off the error.filename field (it just needs to remove \\?\ and \\?\UNC\)
07:52:10  <piscisaureus__>mac os x looks good on my machine, the process.title test fails, and 2 more tests fail because ab is broken on lion
07:52:23  * piscisaureus__leaves
07:54:11  * paddybyersjoined
08:08:24  * piscisaureus__quit (Ping timeout: 245 seconds)
08:10:46  <ryah>[22:14|% 100|+ 639|- 11]: Done OSX
08:11:04  <ryah>[14:58|% 100|+ 644|- 6]: Done Solaris
08:13:03  <ryah>i see.. test-dns.js is failing because the ISP resolver at my parent's house resolves everything...
08:13:13  * piscisaureus__joined
08:13:19  <piscisaureus__>ryah: I'm back
08:13:31  <piscisaureus__>ryah: do we want to get igor's long name stuff in
08:13:34  <ryah>piscisaureus__: good morning
08:13:45  <ryah>piscisaureus__: yes
08:14:07  <piscisaureus__>ok
08:14:09  <ryah>piscisaureus__: up to you - i didn't review it
08:14:13  <ryah>do you think it will create bugs?
08:14:19  <piscisaureus__>ryah: no it looks good
08:14:44  <ryah>k
08:14:48  <piscisaureus__>ryah: we do it in node now for the time being. I flagged down the libuv solution because it would create bugs.
08:22:11  <piscisaureus__>where are exceptions for fs stuff generated
08:22:13  <piscisaureus__>?
08:22:43  <ryah>piscisaureus__: in lib/fs.js
08:23:06  <ryah>nope
08:23:09  <ryah>in src/node_file.cc
08:23:15  <piscisaureus__>hmm. shit
08:24:05  <piscisaureus__>oh hmm it seems that igor fixed that too
08:24:43  <piscisaureus__>igorzi++
08:25:22  <piscisaureus__>that one doesn't look good btw
08:31:13  <piscisaureus__>k. running make test on igor's stuff, probably landing it in 3 minutes
08:37:15  <piscisaureus__>ryah: ok. about EAINONAME. So you replace all occurences of EAINONAME by ENOTFOUND in uv-unix right?
08:38:27  <ryah>piscisaureus__: i forget - what's the issue for this?
08:38:50  <ryah>[24:31|% 100|+ 646|- 4]: Done linux
08:39:06  <piscisaureus__>ryah: I don't know why you did that, but you did it 1 or 2 weeks ago
08:39:47  <ryah>piscisaureus__: hm..
08:39:51  <ryah>piscisaureus__: link?
08:39:59  <piscisaureus__>ryah: yesterday I fixed a c-ares bug where getaddrinfo would basically return a random error
08:40:11  <piscisaureus__>but now test-dns-fail fails for me
08:40:28  <piscisaureus__>because on windows we return EAINOTFOUND where node expects ENOTFOUND
08:41:39  <ryah>test/simple/test-http-dns-fail.js ?
08:41:45  <piscisaureus__>yeah
08:42:42  <piscisaureus__>line 48
08:44:07  <piscisaureus__>ryah: also, it seems that AEI_NONAME is mapped to ENOENT if getaddrinfo is used
08:44:16  <piscisaureus__>in uv-unix
08:44:22  <piscisaureus__>s/AEI/EAI/
08:45:46  * mikealjoined
08:46:15  <ryah>http://github.com/joyent/libuv/commit/2c7e8bb137b96e3e05610e98bfe63bb25df9c7e8
08:47:03  <piscisaureus__>ryah: the thing is that c-ares reports ENOTFOUND
08:47:50  <piscisaureus__>somehow on windows we are using the system resolver for https://github.com/joyent/node/blob/master/test/simple/test-http-dns-fail.js#L34-54
08:47:50  <ryah>piscisaureus__: i think it would be nice to get rid of the EAI* errors
08:47:54  <piscisaureus__>yeah
08:47:56  <ryah>i think that's why i did it
08:48:02  <piscisaureus__>I agree
08:48:13  <piscisaureus__>but if I mirror unix that EAI_NONAME would be mapped to ENOENT
08:48:23  <piscisaureus__>so that test it would still fail on windows
08:49:02  <ryah>why?
08:49:08  <ryah>what is windows reporting?
08:49:22  <piscisaureus__>ryah: EAI_NONAME
08:49:39  <ryah>why not map that to ENOENT?
08:49:55  <piscisaureus__>ryah: that would be in line with unix
08:49:59  <piscisaureus__>ryah: but: https://github.com/joyent/node/blob/master/test/simple/test-http-dns-fail.js#L48
08:51:08  <ryah>piscisaureus__: hm.. there isn't even a UV_ENOTFOUND
08:51:20  <ryah>oh i see. it comes from c-ares..
08:51:21  <piscisaureus__>ryah: sure ... it's a c-ares error
08:51:47  <piscisaureus__>ryah: so somehow in windows the dns resolution goes through the system resolver while on unix it uses cares
08:51:55  <ryah>http shouldnt be using c-ares
08:52:16  <ryah>hm
08:52:21  <ryah>how's that possible?
08:52:48  <piscisaureus__>ryah: I don't know, you are the unix guy :-)
08:52:55  <piscisaureus__>ryah: trying to figure it out
08:53:14  <ryah>oh i see
08:53:15  <piscisaureus__>ryah: dns.js line 34
08:53:27  <ryah>http.request calls dns.lookup which calls cares.getaddrinfo
08:53:34  <piscisaureus__>heh
08:53:49  <ryah>which is actually not cares
08:53:54  <ryah>but just added to the cares binding
08:53:58  <piscisaureus__>ryah: dns.js line 34
08:54:23  <ryah>yeah
08:54:45  <ryah>ok.. so unix isn't going through c-ares
08:54:50  <piscisaureus__>yeha
08:54:52  <ryah>seems you can return ENOENT
08:54:55  <ryah>and it will work
08:54:56  <piscisaureus__>ryah: ok. so I should map to ENOENT
08:54:57  <piscisaureus__>yeah
08:55:31  <piscisaureus__>ryah: I am going to get rid of EAI errors on windows
08:55:59  <piscisaureus__>need new mappings for these:
08:56:00  <piscisaureus__>src/win/getaddrinfo.c: case EAI_FAMILY: return UV_EAIFAMNOSUPPORT;
08:56:01  <piscisaureus__>src/win/getaddrinfo.c: case EAI_NONAME: return UV_EAINONAME;
08:56:01  <piscisaureus__>src/win/getaddrinfo.c: case EAI_SERVICE: return UV_EAISERVICE;
08:56:01  <piscisaureus__>src/win/getaddrinfo.c: case EAI_SOCKTYPE: return UV_EAISOCKTYPE;
08:57:16  * kuebkjoined
08:57:33  <ryah>piscisaureus__: let's just start with fixing this test case and creating an issue for it
08:57:52  <ryah>issue for further removing UV_EAI stuff
08:57:54  <piscisaureus__>k
08:59:11  <ryah>im going to hop in the shower- start putting together the changelog
08:59:17  <ryah>bb in 20
08:59:22  <piscisaureus__>ryah: ok.
08:59:32  <ryah>unixes look okay testing-wise
08:59:39  <piscisaureus__>good
09:05:26  <CIA-95>libuv: Bert Belder v0.6 * rb4698d4 / src/win/fs.c : Strip trailing whitespace - http://git.io/4gqtNg
09:05:26  <CIA-95>libuv: Bert Belder v0.6 * ra2f090f / (include/uv.h src/win/getaddrinfo.c): Win: get rid of UV_EAINONAME, use UV_ENOENT instead - http://git.io/c-0bdA
09:11:01  <piscisaureus__>Shit this breaks binary compat
09:12:26  <CIA-95>libuv: Bert Belder v0.6 * r01f64f6 / (include/uv.h src/win/getaddrinfo.c): Win: get rid of UV_EAINONAME, use UV_ENOENT instead - http://git.io/c38dJw
09:12:28  <eddyb>that must be sweet :)
09:15:19  <piscisaureus__>eddyb: what are you working on btw?
09:15:38  <eddyb>gearbox-node
09:15:43  <piscisaureus__>oh that was it
09:15:47  <piscisaureus__>now I remember
09:15:49  <eddyb>it froze for a month or more
09:15:54  <piscisaureus__>gotta keep track :-)
09:16:01  <eddyb>now I'm trying to remember what I was working on :)
09:17:03  <eddyb>I don't have the Stream/TTY/w/e Wrap interface, so I'm trying to do it with templated functions
09:19:49  <eddyb>damn
09:19:55  <eddyb>too many reinterpret_cast's
09:20:01  <eddyb>that reminds be of an idea I had
09:20:21  <eddyb>to alias ptr_cast (or pointer_cast) to reinterpret_cast
09:21:41  <ryah>piscisaureus__: should i upgrade libuv in node?
09:22:24  <piscisaureus__>ryah: no
09:22:51  <CIA-95>node: Author: Igor Zinkovsky v0.6 * r1f16a7b / (5 files in 3 dirs): Enable long paths on windows - http://git.io/GBsVWw
09:22:52  <CIA-95>node: Bert Belder v0.6 * r5b014e0 / src/node_file.cc : Fs: use strncmp instead of memcmp for comparing strings - http://git.io/8vKFNQ
09:22:52  <CIA-95>node: Bert Belder v0.6 * ra056ef1 / (3 files in 2 dirs): Upgrade libuv to 01f64f6 - http://git.io/Ktf4Ng
09:24:52  * ryahtests again
09:25:10  <piscisaureus__>ryah: sorry...
09:27:38  <eddyb>uhm...
09:27:59  <eddyb>this is C++... yet it's still using strncmp?
09:28:26  <piscisaureus__>eddyb: then what should it be using?
09:28:39  <piscisaureus__>it's C/C++
09:28:39  <eddyb>a string class?
09:28:43  <piscisaureus__>nah
09:28:45  <piscisaureus__>nope
09:28:55  <eddyb>that reminds me I should make my String class inherit string
09:29:18  * ryahrolls his eyes
09:29:20  <eddyb>std::string, that is
09:30:34  <eddyb>so I would only need to write operator char*() and operator v8::Handle<v8::String> (hopefully I can make it faster, too)
09:31:04  <CIA-95>node: Bert Belder v0.6 * rb480cfa / test/pummel/testcfg.py : Make the pummel test runner not crash on windows when a directory is locked - http://git.io/FDvYtg
09:31:05  <piscisaureus__>^-- I wanted to get that in a long time. No need to retest, ryah
09:31:23  <piscisaureus__>windows 7: [04:14|% 100|+ 317|- 10]
09:31:39  <piscisaureus__>2 failures due to ISP not supporting ipv6
09:31:49  <ryah>col
09:32:01  <ryah>let's get an RC
09:32:05  <piscisaureus__>1 spurious error
09:32:06  <piscisaureus__>yeah
09:32:25  <piscisaureus__>I guess we have to write that doc first right?
09:32:34  <ryah>yes
09:32:47  * ryahgets another beer
09:32:47  <ryah>:)
09:32:50  <piscisaureus__>:-)
09:33:32  <piscisaureus__>ryah: http://beta.etherpad.org/p/v0.6.3
09:38:28  <piscisaureus__>crap it sucks
09:39:21  <ryah>piscisaureus__: the way i do it is git log v0.6.2^..HEAD | vim -
09:39:27  <ryah>then :sn Changelog
09:39:40  <piscisaureus__>ryah: that's what I did :-)
09:40:05  <piscisaureus__>same for libuv: git log -n1 v0.6.2 -- deps/uv
09:40:25  <piscisaureus__>then iin libuv - git log 01fbe7b..v0.6
09:42:47  * igorzijoined
09:43:41  <igorzi>piscisaureus_: thanks for landing that patch (and fixing memcmp)
09:44:02  <igorzi>how's the release looking?
09:45:26  <ryah>waiting for the changelog :)
09:45:41  <piscisaureus__>Looking good, writing the changelog
09:46:39  <ryah>=== release test-fs-realpath ===
09:46:39  <ryah>Path: simple/test-fs-realpath
09:46:40  <ryah>node.js:201
09:46:40  <ryah> throw e; // process.nextTick error, or 'error' event on first tick
09:46:41  <ryah> ^
09:46:44  <ryah>AssertionError: "/" == "C:\\"
09:46:47  <ryah> at Object.<anonymous> (C:\Users\ryan\node\test\simple\test-fs-realpath.js:46
09:46:49  <ryah>8:8)
09:46:52  <ryah> at Module._compile (module.js:432:26)
09:46:54  <ryah> at Object..js (module.js:450:10)
09:46:56  <ryah> at Module.load (module.js:351:31)
09:46:57  <piscisaureus__>ryah: known problem. Has never worked
09:46:59  <ryah> at Function._load (module.js:310:12)
09:47:01  <ryah> at Array.0 (module.js:470:10)
09:47:04  <piscisaureus__>ryah: windows has not realpath support atm
09:47:04  <ryah> at EventEmitter._tickCallback (node.js:192:40)
09:47:07  <ryah>Command: C:\Users\ryan\node\Release\node.exe C:\Users\ryan\node\test\simple\test
09:47:09  <ryah>-fs-realpath.js
09:47:11  <ryah>^-- is this known?
09:47:14  <ryah>k
09:49:25  <ryah>sun: [12:32|% 100|+ 649|- 5]: Done
09:50:11  <ryah>win7: [08:04|% 100|+ 317|- 10]: Done
09:56:38  <piscisaureus__>Almost done
09:58:02  <ryah>http://nodejs.org/dist/v0.6.3/node-v0.6.3-rc3.msi
09:58:11  <ryah>^-- i preemptively made a RC
09:58:21  <ryah>in case anyone wants to test
10:00:07  <piscisaureus__>ryah: https://gist.github.com/123a435d55c00ebb5fe4
10:00:35  <ryah>piscisaureus__: looks good
10:00:44  <ryah>oh, don't put my name in there
10:00:54  <piscisaureus__>ok
10:00:55  <piscisaureus__>why?
10:01:17  <ryah>*shrug* never put it in before :)
10:03:36  <piscisaureus__>where is NODE_VERSION_MINOR defined again?
10:03:57  <piscisaureus__>hmm nvm got it
10:11:34  <CIA-95>node: Bert Belder v0.6 * r2335c4b / ChangeLog : Strip trailing whitespace in changelog - http://git.io/B-oh9A
10:11:34  <CIA-95>node: Bert Belder v0.6 * rfdcfbd6 / (ChangeLog doc/index.html src/node_version.h): Bump version to v0.6.3 - http://git.io/aSTfjA
10:13:26  <piscisaureus__>ryah: okay what next?
10:13:46  <piscisaureus__>30 minutes to release
10:17:43  <ryah>piscisaureus__: but the rc
10:17:45  <ryah>cut
10:18:07  <ryah>piscisaureus__: hm
10:18:15  <ryah>piscisaureus__: you should let me review that commit
10:18:35  <ryah>you missed updating version in doc/template.html
10:19:20  <ryah>piscisaureus__: also add the sha1 tag for 0.6.2 in the changelog
10:19:32  <piscisaureus__>ryah: ah I forgot the sha1
10:19:34  <ryah>piscisaureus__: update the commit and force push the branch
10:20:52  <piscisaureus__>there have not been any sha1s in the changelog since 0.6.0
10:21:16  <ryah>oh yeah, that's bad
10:21:17  <piscisaureus__>oh that is only 0.6.1
10:21:30  <ryah>well whatever
10:21:55  <piscisaureus__>adding it
10:22:22  <ryah>piscisaureus__: are you on a mac?
10:22:33  <piscisaureus__>not yet
10:22:42  <piscisaureus__>can reboot
10:22:47  <piscisaureus__>let me first update this commiit
10:23:53  <CIA-95>node: Bert Belder v0.6 * rb159c6d / (4 files in 3 dirs): Bump version to v0.6.3 - http://git.io/XEyEXg
10:24:04  <piscisaureus__>ryah: ^-- that better?
10:24:26  <ryah>yes
10:25:47  <ryah>piscisaureus__: reboot to mac and do "make dist-upload"
10:26:17  <ryah>or should i just do it?
10:26:23  <ryah>let me just do it..
10:26:24  * piscisaureus__quit (Quit: ~ Trillian Astra - www.trillian.im ~)
10:26:41  * ryahwaits
10:29:13  * piscisaureujoined
10:29:25  <piscisaureu>ryah: make dist-upload fails for me on mac
10:29:34  <ryah>piscisaureu: how?
10:29:40  <piscisaureu>git archive --format=tar --prefix=node-v0.6.3/ HEAD | tar xf -
10:29:40  <piscisaureu>mkdir -p node-v0.6.3/doc
10:29:40  <piscisaureu>cp doc/node.1 node-v0.6.3/doc/node.1
10:29:40  <piscisaureu>cp -r out/doc/api node-v0.6.3/doc/api
10:29:40  <piscisaureu>cp: out/doc/api: No such file or directory
10:29:40  <piscisaureu>make: *** [node-v0.6.3.tar.gz] Error 1
10:30:05  <ryah>hm
10:30:09  <piscisaureu>should I run make doc first?
10:30:18  <ryah>yeah try that
10:30:34  <piscisaureu>yeah works
10:31:41  <ryah>piscisaureu: prepare blog.nodejs.org post
10:31:43  <piscisaureu>ryah: I hope the pkg works
10:31:47  <ryah>piscisaureu: i'll test
10:31:52  <eddyb>yay, dual monitor again :)
10:32:03  <ryah>piscisaureu: just to make sure - you're running this on v0.6.3 b159c6d ?
10:32:21  <piscisaureu>ryah: yep
10:32:52  <ryah>piscisaureu: pkg works
10:33:13  <ryah>piscisaureu: im going to now build the msi
10:33:21  <ryah>(from the tarball)
10:33:23  <piscisaureu>cool.
10:33:29  <piscisaureu>also build the pdb
10:34:31  * igorziquit (Ping timeout: 265 seconds)
10:36:12  <ryah>piscisaureu: did you git tag the release?
10:41:18  <piscisaureu>ryah: not yet, will do
10:41:20  <piscisaureu>http://blog.nodejs.org/2011/11/25/node-v0-6-3-stable/
10:41:33  <piscisaureu>ryah: or - does that need to be done first?
10:42:46  <ryah>piscisaureu: i can't see that
10:46:18  * igorzijoined
10:47:03  <igorzi>[08:29|% 100|+ 314|- 12]: Done <- win7 test-all with http://nodejs.org/dist/v0.6.3/node-v0.6.3-rc3.msi
10:47:26  <ryah>ok - just uploaded the real .msi
10:48:25  <ryah>piscisaureu: can you save that blog post as a draft so i can check it out
10:48:57  <piscisaureu>ryah: I did :-/
10:49:12  <piscisaureu>http://blog.nodejs.org/?p=365&preview=true
10:49:37  <ryah>http://nodejs.org/docs/v0.6.3/ <--
10:49:47  <ryah>http://nodejs.org/docs/v0.6.3/api/ <--
10:49:49  <ryah>both 404
10:50:08  <ryah>scp -r out/doc [email protected]:~/web/nodejs.org/docs/v0.6.3
10:50:25  <piscisaureu>ryah: maybe it went wrong because I didn't tag before make dist-upload
10:50:33  <piscisaureu>I don't know what it does then
10:50:58  <piscisaureu>ah
10:51:05  <piscisaureu>it is website-upload right
10:51:25  <ryah>you have to manually upload the doc
10:52:21  <piscisaureu>happening
10:53:03  <ryah>(before the release i go through the blog post and click all the links)
10:53:32  <ryah>piscisaureu: okay
10:53:38  <ryah>piscisaureu: post the blog
10:53:40  <igorzi>also, i see that msi and node.exe (installed by the msi) are signed.. so that's good
10:54:22  <piscisaureu>ryah: links all work
10:55:21  <piscisaureu>ryah: tag is also there
10:55:26  <eddyb>btw
10:55:37  <ryah>piscisaureu: also make a "working on v0.6.4 now"
10:55:40  <ryah>commit
10:55:48  <piscisaureu>ryah: yeah
10:55:51  <eddyb>does automated testing work with the gyp version?
10:56:08  <piscisaureu>ryah: do you use git push --tags?
10:56:14  <ryah>yes
10:56:21  <ryah>eddyb: yes
10:56:30  <eddyb>oh, yeah... gyp is default, too
10:56:32  <ryah>eddyb: well - define automated teting?
10:56:32  <piscisaureu>ryah: there are a lot of bogus tags in github
10:56:42  <ryah>piscisaureu: i know - i dont know how to remove them
10:56:44  <piscisaureu>ryah: I removed them at some point but now they are all back
10:56:45  <eddyb>ryah: make test or something like that
10:56:55  <eddyb>test-all, I think
10:56:57  <piscisaureu>https://github.com/joyent/node/tags?after=v0.0.1
10:57:04  <ryah>eddyb: yes - it works with the gyp build
10:57:29  <ryah>piscisaureu: twitter update, mailing list post
10:57:37  <eddyb>I should start kdiff3-ing some files
10:57:38  <piscisaureu>ryah: yeah
10:57:46  <piscisaureu>ryah: rebooting again firss
10:57:54  * piscisaureuquit (Quit: Leaving.)
10:59:12  <igorzi>i'm off. good night.
10:59:16  <ryah>igorzi: night!
10:59:34  * igorziquit (Quit: Page closed)
10:59:37  * piscisaureus__joined
11:00:47  <ryah>piscisaureus__: something liek "#nodejs v0.6.3 released with npm included! [link]"
11:01:10  <piscisaureus__>ryah: I was doing exactly that :-)
11:03:08  <ryah>piscisaureus__: also "make website-upload"
11:03:15  <piscisaureus__>ryah: already done
11:03:16  <ryah>oh that's already done
11:03:19  <ryah>ok
11:03:23  <piscisaureus__>next stop: mailing list
11:04:49  <ryah>piscisaureus__: don't forget the "now owrking on v0.6.4" commit.
11:04:53  <ryah>im off to bed
11:05:03  <piscisaureus__>ryah: will do. I will also merge v0.6 back to maser
11:05:04  <ryah>piscisaureus__: good job :)
11:05:05  <piscisaureus__>*master
11:05:12  <piscisaureus__>ryah: sleep tight
11:05:17  <ryah>cool
11:05:20  <ryah>ciao
11:10:25  <piscisaureus__>crap
11:10:25  <piscisaureus__>npm ERR! Error: ENOENT, no such file or directory 'C:\Program Files (x86)\nodejs
11:10:26  <piscisaureus__>\node_modules\npm\man\man1'
11:12:24  <eddyb>https://github.com/eddyb/gearbox-node/blob/master/node.gyp#L132-136
11:12:34  <eddyb>someone should add that to node
11:12:54  <piscisaureus__>yeah
11:13:08  <eddyb>otherwise it uses local v8 even when it's set to shared
11:17:24  <piscisaureus__>eddyb: https://github.com/joyent/node/issues/2187
11:19:46  <eddyb>piscisaureus__: thanks, that's cool :)
11:28:41  <eddyb>piscisaureus__: also, commented on it
11:28:50  <piscisaureus__>k
11:29:35  <eddyb>and also edited it with the link
11:29:50  <eddyb>should be easy to do :)
11:35:15  <CIA-95>node: Bert Belder master * rcffd0bb / (171 files in 62 dirs):
11:35:15  <CIA-95>node: Merge branch 'v0.6'
11:35:15  <CIA-95>node: Conflicts:
11:35:15  <CIA-95>node: Makefile
11:35:15  <CIA-95>node: src/node_version.h
11:35:16  <CIA-95>node: test/simple/test-http-response-no-headers.js (+17 more commits...) - http://git.io/jeBSOQ
11:35:50  * bnoordhuisjoined
11:35:55  <eddyb>did I just find a bug?
11:35:57  <eddyb>configure options: {'shared_v8': True, 'openssl_includes': None, 'without_snapshot': None, 'shared_v8_includes': None, 'shared_v8_libpath': None, 'openssl_libpath': None, 'shared_cares_includes': None, 'dest_cpu': 'x64', 'without_ssl': None, 'shared_cares': None, 'prefix': None, 'no_ssl2': None, 'debug': None, 'shared_cares_libpath': None, 'shared_v8_libname': None, 'gdb': None, 'with_dtrace': None}
11:36:02  <eddyb>['-f', 'make', '/home/eddy/Projects/gearbox-node/node.gyp', '-I', '/home/eddy/Projects/gearbox-node/common.gypi', '-I', '/home/eddy/Projects/gearbox-node/options.gypi', '--depth=.', '--generator-output', '/home/eddy/Projects/gearbox-node/out', '-Goutput_dir=/home/eddy/Projects/gearbox-node/out', '-Dtarget_arch=ia32', '-Dcomponent=static_library', '-Dlibrary=static_library']
11:36:16  <eddyb>target_arch is ia32
11:36:29  <eddyb>even if I have dest_cpu set to x64
11:37:03  <piscisaureus__>could be, I don't know
11:37:12  <bnoordhuis>i'm pretty sure we don't honour DEST_OS right now
11:37:28  <piscisaureus__>the gyp build on mac always crashes for me on mac so yeah probably
11:37:37  <piscisaureus__>eddyb: please file a bug
11:38:34  <eddyb>will do
11:39:15  <eddyb>piscisaureus__: can you mark https://github.com/joyent/node/issues/2116 as a duplicate, or merge the two?
11:40:52  <piscisaureus__>yup
11:44:51  <CIA-95>node: Bert Belder v0.6 * r3a16064 / src/node_version.h : Now working on v0.6.4 - http://git.io/xWDbbg
11:45:54  <eddyb>https://github.com/joyent/node/issues/2189
11:47:13  <piscisaureus__>k thnx
11:47:41  <CIA-95>libuv: Bert Belder master * r35fa2a6 / (6 files in 2 dirs): Merge branch 'v0.6' (+8 more commits...) - http://git.io/8opqyQ
11:49:19  <piscisaureus__>eddyb: kewl
14:31:51  <piscisaureus__>https on windows is extremely slow
14:33:20  <bnoordhuis>piscisaureus__: define 'extremely slow'
14:34:01  <piscisaureus__>10 kb per second
14:34:19  <piscisaureus__>bnoordhuis: it could be the https server though
14:34:34  <piscisaureus__>bnoordhuis: can you try https://registry.npmjs.org/cloud9/-/cloud9-0.5.1.tgz
14:34:48  <bnoordhuis>piscisaureus__: with what?
14:35:28  <piscisaureus__>bnoordhuis: https://gist.github.com/1393656
14:36:29  <bnoordhuis>piscisaureus__: ~175K/s
14:36:56  <eddyb>it takes ages
14:36:59  <bnoordhuis>well... it's jittery
14:37:04  <eddyb>to download 100KB
14:37:19  <eddyb>actually, I still have 60KB
14:37:25  <eddyb>and this is on linux
14:37:32  <bnoordhuis>there's a ridiculous amount of tx traffic
14:37:36  <piscisaureus__>for me it is also super slow
14:37:43  <piscisaureus__>but not when I use normal http
14:37:45  <bnoordhuis>i think we need smarter session caching
14:38:05  <bnoordhuis>piscisaureus__: 1m2.240s <- that's how long it took
14:38:27  <piscisaureus__>After 1 minute I have downloaded ~1mb
14:38:32  <piscisaureus__>so it will take 25 minutes
14:39:09  <bnoordhuis>odd - tls is mostly independent from the network stak
14:39:11  <bnoordhuis>*stack
14:39:15  <bnoordhuis>fscking keyboard
14:39:17  <piscisaureus__>yeah
14:39:24  <piscisaureus__>could also be the isp or something
14:39:33  <piscisaureus__>where can I download a big file over https?
14:39:39  <bnoordhuis>github?
14:39:42  <piscisaureus__>:-)
14:39:43  <piscisaureus__>good one
14:40:50  <eddyb>60KB
14:40:57  <eddyb>it's been like that for minutes
14:41:07  <eddyb>3 minutes, that is
14:42:36  <piscisaureus__>github is also extremely slow
14:42:44  <piscisaureus__>https://nodeload.github.com/joyent/node/tarball/master
14:44:15  <piscisaureus__>hmpf what's going on
14:44:30  <bnoordhuis>piscisaureus__: that tarball downloaded like in 10s for me
14:44:32  <piscisaureus__>node is not using any memory
14:44:37  <piscisaureus__>bnoordhuis: in node?
14:44:39  <bnoordhuis>yes
14:44:48  <piscisaureus__>bnoordhuis: this is definitely a windows issue then
14:44:56  <piscisaureus__>normal https is also very fast for me
14:45:09  <bnoordhuis>normal as in non-node?
14:45:24  <eddyb>happens on linux, too
14:45:25  <piscisaureus__>normal as in unencryped
14:45:38  <eddyb>altho, it's 0.5-something
14:45:50  <bnoordhuis>eddyb: upgrade :)
14:46:06  <eddyb>bnoordhuis: I built the package myself
14:46:18  <bnoordhuis>piscisaureus__: 'normal https' vs 'unencrypted'? typo?
14:46:24  <piscisaureus__>ah type
14:46:27  <piscisaureus__>typo
14:46:28  <eddyb>so I need to go into OBS settings and tweak it
14:47:05  <bnoordhuis>piscisaureus__: talking about tls, can you review this PR -> https://github.com/joyent/node/pull/1983
14:48:19  <piscisaureus__>using non-overlapped tcp reads doesn't change it
14:48:20  <piscisaureus__>that's good
14:50:44  <piscisaureus__>bnoordhuis: do you have a windows machine?
14:50:56  <bnoordhuis>piscisaureus__: no, but i have a windows vm
14:52:05  <piscisaureus__>hmm
14:52:29  <piscisaureus__>most of the time libuv is blocked on GetQueuedCompletionStatusEx
14:55:22  <piscisaureus__>so nice to be able to hack your networking stack :-)
15:01:27  <piscisaureus__>bnoordhuis: re 1983 - did ryah ack that
15:01:32  <piscisaureus__>bnoordhuis: ?
15:01:44  <piscisaureus__>bnoordhuis: since this makes the api less consistent
15:04:26  <bnoordhuis>piscisaureus__: less? i think more consistent?
15:05:04  <piscisaureus__>bnoordhuis: well net.createConnection doesn't take an options
15:05:18  <piscisaureus__>bnoordhuis: have you tried your windows vm yet?
15:05:51  <bnoordhuis>piscisaureus__: no, i'm working on three other things already :/
15:06:31  <piscisaureus__>this is really super bad
15:06:44  <piscisaureus__>because people are going to try npm and they will think it's not working
15:28:19  <piscisaureus__>bnoordhuis: somehow the problem is less when I block the event loop for a while on every package
15:28:33  <piscisaureus__>with for (var i = 0; i < 100000000; i++);
15:28:44  <piscisaureus__>s/package/packet/
15:35:28  <bnoordhuis>sizeof(struct stat) = 96
15:35:28  <bnoordhuis>sizeof(struct stat) = 108
15:35:28  <bnoordhuis>^ that's why node is crashing on os x >:(
15:35:40  <piscisaureus__>bnoordhuis: the gyp build is also crashing
15:35:52  <piscisaureus__>bnoordhuis: because v8 gets built for ia32 indistriminately
15:36:03  <bnoordhuis>imma fix that
15:45:31  <bnoordhuis>we're over 11K watchers now! https://github.com/popular/watched
15:51:10  * AvianFlujoined
15:54:10  <CIA-95>node: Ben Noordhuis master * r1cf13bc / common.gypi :
15:54:10  <CIA-95>node: build: compile with -D__DARWIN_64_BIT_INO_T on OS X
15:54:10  <CIA-95>node: Fixes a struct stat size mismatch on 64 bits machines that made Node crash with
15:54:10  <CIA-95>node: a EXC_BAD_ACCESS on startup.
15:54:10  <CIA-95>node: Fixes #2061 for gyp builds. Solution proposed by Paddy Byers. - http://git.io/5doHwQ
16:18:57  <piscisaureus__>okay. read short-circuiting is causing the https shit
16:19:58  <piscisaureus__>we need to fix this real soon
16:20:35  <piscisaureus__>the problem is that ssl processing is deferred to the next tick
16:21:25  <piscisaureus__>libuv always drains the read buffer real fast
16:21:38  <piscisaureus__>until it is really empty and then it starts processing
16:21:49  <piscisaureus__>so the read buffer fills up and throttlink kicks in
16:22:26  * mikealquit (Quit: Leaving.)
16:29:08  * paddybyerspart
16:33:13  * piscisaureus__quit (Quit: ~ Trillian Astra - www.trillian.im ~)
16:39:28  <bnoordhuis>piscisaureus_: https://github.com/joyent/node/pull/1983 <- if you object to it, can you post your objections in the issue?
16:43:52  * paddybyersjoined
16:55:31  * kuebkquit (Quit: Leaving.)
18:37:04  * piscisaureus__joined
18:38:50  * igorzijoined
18:39:44  <igorzi>piscisaureus_: hey
18:40:20  <igorzi>piscisaureus_: what'sup with https?
18:43:46  <igorzi>piscisaureus__: --^
18:53:48  <piscisaureus__>igorzi: see email
18:54:17  <piscisaureus__>igorzi: try `npm install cloud9`
18:54:42  <piscisaureus__>probably for me it's worse because eu-us latency is much higher that us-us latency
18:56:06  <igorzi>piscisaureus__: just read it.. do you already have a fix in mind?
18:57:13  <piscisaureus__>igorzi: yeah: the "completed request queue" needs to alternate
18:57:29  <piscisaureus__>e.g. we never process a request that has short-circuited on the same tick
18:57:30  <piscisaureus__>I think
18:58:16  <piscisaureus__>it's a little difficult because we also have an "endgame" queue
18:58:43  <piscisaureus__>I wanted to make that the same queue for a while, but that requires requests and handles to have a common base class
18:58:49  <piscisaureus__>I never got around doing that
19:00:56  <igorzi>piscisaureus__: i remember you dealing with something similar with udp? (short-circuited reads starving prepare and check?)
19:01:30  <piscisaureus__>igorzi: yes, I never dealt with that
19:01:43  <piscisaureus__>igorzi: it causes udp packet storm benchmarks to time out
19:02:03  <piscisaureus__>because libuv always short circuits and never makes the timer callback that is supposed to end the benchmark
19:02:10  <igorzi>piscisaureus__: but fundamentally - it's the same issue, right?
19:02:26  <piscisaureus__>igorzi: yeah
19:03:07  <igorzi>piscisaureus__: so you're going to put some "shortcircuited" flag on a request?
19:03:16  <piscisaureus__>igorzi: no
19:03:38  <igorzi>piscisaureus__: different queue?
19:03:53  <piscisaureus__>igorzi: what I want to do is this: before calling all the requests we make a copy of the pending_reqs queue
19:04:00  <piscisaureus__>and clear out the original one
19:04:11  <piscisaureus__>and then we only process the requests in the queue copy
19:04:25  <piscisaureus__>it is really easy - since it is a linked list we just have to copy the head of the queue
19:05:37  <piscisaureus__>igorzi: let's not do it with the endgame queue for the time being
19:05:55  <piscisaureus__>just make sure it is fully drained before calling uv_poll again
19:06:17  <piscisaureus__>igorzi: uv_poll has a `block` argument
19:06:30  <piscisaureus__>we make that false if there are short-circuited requests in the queue
19:07:35  <igorzi>piscisaureus__: one thing i don't fully follow - is where do these read requests come from? with tcp we get a 0-read read req, and then do blocking reads.. so after we drain the buffer, where do new reqs come from?
19:08:58  <piscisaureus__>well, the buffer gets replenished very quickly
19:09:31  <piscisaureus__>igorzi: node does some processing in the read callback, after that is done there already is new data in the read buffer
19:09:58  <piscisaureus__>igorzi: I don't know if you can reproduce the issue
19:10:46  <piscisaureus__>igorzi: if you can, try disabling othe SetHandleCompletionNotificationModes call and you'll see it all works fine
19:11:27  <igorzi>piscisaureus__: ok.. so then eventually we drain everything, post another 0-read, which again short-circuits? and prepare+check never get to run?
19:11:47  <piscisaureus__>igorzi: yes
19:12:56  <piscisaureus__>igorzi: K
19:13:10  <igorzi>piscisaureus__: cool.. so your fix will take a snapshot of reqs, and just process them without processing any new reqs that get added, correct?
19:14:01  <piscisaureus__>igorzi: I use this to reproduce it: https://gist.github.com/1393656
19:14:05  <piscisaureus__>igorzi: yes, correcto
19:14:47  <piscisaureus__>igorzi: it will kind of eliminate the benefit of short-circuiting when there is only 1 connection
19:14:58  <piscisaureus__>igorzi: but for multiple connections it'll still work
19:15:38  <piscisaureus__>igorzi: it also works for accepts; since we queue 32 requests, 32 accepts can short-circuit per loop iteration
19:17:15  <igorzi>piscisaureus__: i've never seend accepts short-circuiting (at least with my lsp)
19:17:59  <piscisaureus__>igorzi: I am not so sure either
19:18:12  <piscisaureus__>igorzi: well, in theory it could :-)
19:18:45  <igorzi>piscisaureus__: right
19:20:06  <igorzi>piscisaureus__: does processing prepare+check need to go through IOCP?
19:20:22  <piscisaureus__>igorzi: not necessarily
19:20:34  <piscisaureus__>igorzi: but under some circumstances it can be bad not to do that
19:21:07  <piscisaureus__>I think
19:21:26  <piscisaureus__>let me think about that for a while
19:21:28  <igorzi>piscisaureus__: cause it would be cool to not starve prepare+check, but still not go to IOCP if we have some short-circuited reqs to process
19:21:29  <piscisaureus__>hmm
19:21:47  <piscisaureus__>igorzi: the risk is that, suppose you have a connection that you can write to really fast
19:21:51  <piscisaureus__>so it short-circuits all the time
19:22:05  <piscisaureus__>then you will only spend time writing to that connection
19:22:32  <piscisaureus__>other connections never get any attention because we never query the completion port
19:23:27  <igorzi>piscisaureus__: hmm.. maybe impose a threshold on how many reqs we can process without checking in IOCP?
19:23:50  <piscisaureus__>igorzi: yes that could also work
19:24:14  <piscisaureus__>but what would that number be?
19:25:31  <piscisaureus__>igorzi: btw are you off today?
19:25:38  <igorzi>piscisaureus__: yep
19:26:06  <piscisaureus__>igorzi: then you should go have fun :-)
19:26:16  <igorzi>it seems that number could be pretty large.. it's just there to guard against this type of infinite write scenario
19:26:30  <igorzi>piscisaureus__: yeah.. i'm getting off in 5 min :)
19:27:00  <piscisaureus__>igorzi: yeah, but let's do some experiments with that later and get this fixed first :-)
19:27:27  <igorzi>piscisaureus__: yep, sounds good
19:29:57  <igorzi>piscisaureus__: bye, ttyl
19:30:38  * igorziquit (Quit: Page closed)
19:31:01  <piscisaureus__>igorzi: byte. have fun
19:39:00  * felixgequit (Quit: felixge)
20:17:34  * mralephjoined
20:52:10  * felixgejoined
20:52:10  * felixgequit (Changing host)
20:52:10  * felixgejoined
20:52:14  * felixgequit (Client Quit)
21:07:50  <bnoordhuis>https://github.com/popular/watched <- woo!
21:14:22  <piscisaureus__>bnoordhuis: yup, no-one to beat!
21:14:32  <piscisaureus__>bnoordhuis: from here we can only lose
21:14:41  <bnoordhuis>yeah, that's the downside :/
21:20:32  <piscisaureus__>bnoordhuis: I'm off. time for a drink
21:21:11  <bnoordhuis>piscisaureus__: have fun
21:34:11  <piscisaureus__>bnoordhuis: one more thing
21:34:13  <piscisaureus__>http://groups.google.com/group/nodejs/msg/94830f69a2392d7e
21:34:20  <piscisaureus__>bnoordhuis: I agree.
21:34:29  <bnoordhuis>meh
21:34:35  <bnoordhuis>i don't
21:34:57  <bnoordhuis>`make install` also installs node-waf and i don't hear people complaining about that
21:35:21  <piscisaureus__>bnoordhuis: people don't install node-waf separately
21:35:39  <piscisaureus__>bnoordhuis: I think we could just have 'make install-bare'
21:37:13  <bnoordhuis>piscisaureus__: i don't feel strongly about but i think it's just people bitching
21:37:29  <bnoordhuis>19 of 20 people will be happy that npm is finally bundled with node now
21:41:25  <piscisaureus__>[email protected]
21:42:01  <piscisaureus__>bnoordhuis: we should just add make install-nonpm and remove that option later
21:42:24  <piscisaureus__>bnoordhuis: I think most people are angry because they already have a working npm installed and they don't want to fuck that up
21:42:29  <piscisaureus__>this will obviously change
21:42:32  <piscisaureus__>over time
21:43:17  <bnoordhuis>piscisaureus__: well, it's too late for 0.6.3 and 0.6.4 is going to fix that, right?
21:43:57  <piscisaureus__>bnoordhuis: it should
21:44:04  <piscisaureus__>bnoordhuis: we should get it out on monday
21:44:07  <piscisaureus__>or tuesday
21:44:16  <piscisaureus__>fix just these few issues
21:45:10  <bnoordhuis>piscisaureus__: right, so no need for a 'non npm' install option
21:58:04  * piscisaureus__quit (Quit: ~ Trillian Astra - www.trillian.im ~)
22:00:08  * piscisaureus__joined
22:00:09  * piscisaureus__quit (Remote host closed the connection)
22:46:33  * mralephquit (Quit: Leaving.)
22:47:57  * mralephjoined
22:55:25  * mralephquit (Quit: Leaving.)
23:43:25  * AvianFluquit (Quit: Leaving)
23:58:16  * piscisaureus__joined
23:58:49  * piscisaureus_quit (Read error: Connection reset by peer)