00:02:06  <igorzi>no mention of fs watcher? waiting for kqueue implementation?
00:02:12  <ryah>igorzi: yes
00:23:10  * piscisaureusjoined
00:40:39  <igorzi>ryah: what are the exact semantics of uv_read2_cb? when pending is not UV_UNKNOWN_HANDLE, what is expected in nread and buf?
00:42:35  * erickt_quit (Ping timeout: 256 seconds)
00:49:20  <igorzi>ryah: it's not clear from the existing test-ipc. my assumption is that pending and nread are completely separate. if nread is non-zero, that doesn't indicate that pending is not UV_UNKNOWN_HANDLE, right?
00:55:57  * piscisaureusquit (Ping timeout: 258 seconds)
01:01:15  <ryah>igorzi: corerct
01:01:33  <ryah>igorzi: there can be a handle and data - just a handle - or jsut data
01:01:44  <ryah>or neither - in the case of EOF
01:13:11  <igorzi>ryah: ok.. on windows impl, i think we'll never have handle and data. i just wanted to make sure that the test didn't assume that you'll have both at the same time
01:18:52  <ryah>igorzi: hm - why can't you send handle and data at the same time?
01:51:27  <igorzi>ryah: i'm trying to avoid allocating temp buffers and copying
01:51:52  <igorzi>ryah: i think we could send both if we had free_cb (in addition to alloc_cb)
02:03:59  * brsonquit (Ping timeout: 248 seconds)
02:11:27  * isaacsquit (Quit: isaacs)
02:12:21  * luxigo_quit (Read error: Connection reset by peer)
02:25:19  * isaacsjoined
02:26:22  * isaacsquit (Client Quit)
02:52:29  * erickt_joined
03:17:11  <igorzi>ryah: nm.. i'll be able to send both.
03:30:42  <erickt_>I noticed that there's a slight difference between src/win/tcp.c:uv_tcp_set_socket and src/win/udp.c:uv_udp_set_socket: https://gist.github.com/1255548
03:30:57  <erickt_>Was this on purpose? Or should they be synced up?
03:32:30  <erickt_>igorzi: it looks like you made this change to tcp.c
03:33:21  <erickt_>In commit e67b9a37. The commit message is "windows: fix issue with SetFileCompletionNotificationModes on 64bit xp"
03:56:33  * isaacsjoined
04:37:15  * erickt_quit (Quit: erickt_)
04:40:08  * felixgejoined
04:40:08  * felixgequit (Changing host)
04:40:09  * felixgejoined
05:16:02  <indutny>can I install SmartOS on vmware?
05:21:41  <ryah>indutny: yeah
05:21:51  <indutny>cool
05:22:02  <indutny>I'll try building uv_uptime on it, though
05:23:06  <indutny>ryah: btw, can I cache values in static variables in libuv or is at a bad practice?
05:23:26  <indutny>ryah: getting "System Up Time" titleIndex is very slow and allocating 70kb of memory
05:26:25  <ryah>:/
05:26:38  <indutny>yeah, fcking winap
05:26:42  <indutny>winapi*
05:26:50  <ryah>indutny: why cache?
05:27:14  <ryah>indutny: i dont think node users call this often
05:27:16  <indutny>ryah: because it takes about 400ms to request "System Up Time" titleIndex from registry
05:27:20  <indutny>ok
05:27:26  <indutny>what about 70kb allocation?
05:28:25  <ryah>70kb on libuv's side?
05:28:38  <indutny>yes
05:28:43  <indutny>malloc
05:31:30  <ryah>*shrug* i dont see how a cache fixes that
05:32:53  <ryah>that just hides the problem - just put it in the user's face
05:33:55  <DrPizza>what is someone doing that is causing them to query the uptime regularly
05:35:57  <indutny>DrPizza: calling uv_uptime() ? :)
05:36:00  <indutny>regularly
05:36:27  <indutny>ryah: I can use const value for TitleIndex
05:36:48  <indutny>it's well known, but there're are no docs about it.
05:36:53  <indutny>so probably it may change
05:36:56  <indutny>674
05:37:01  <indutny>DrPizza: What do you think?
05:37:34  <DrPizza>see if 't sthesame on all supported OSes
05:37:42  <DrPizza>but the way i read the docs
05:37:45  <DrPizza>009 is always available
05:37:50  <indutny>yes
05:37:53  <indutny>of course
05:38:08  <indutny>that's why I'm requesting it :)
05:38:17  <DrPizza>oh you were asking earlier what I meant
05:38:27  <indutny>aah
05:38:36  <indutny>sorry, don't get it anyway :)
05:38:45  <DrPizza>if 009 is always available
05:38:48  <DrPizza>you don't need to localize
05:39:06  <indutny>DrPizza: but how do I get "System Up Time" titleId ?
05:39:13  <DrPizza>which means you can just use PDH
05:39:19  <DrPizza>and just request system up time
05:39:34  <indutny>DrPizza: http://msdn.microsoft.com/en-us/library/aa372536%28VS.85%29.aspx
05:39:49  <indutny>why do you think they added this function
05:40:03  <indutny>This function provides a language-neutral way
05:40:09  <indutny>to add counters to a query
05:40:21  <indutny>Minimum supported client Windows Vista
05:42:33  <indutny>DrPizza: so what?
05:43:02  <DrPizza>no idea
05:43:49  <DrPizza>but like
05:43:53  <DrPizza>if you don't specify a language
05:43:58  <DrPizza>it uses english anyway
05:44:00  <DrPizza>so can't you just
05:44:01  <DrPizza>use english?
05:44:15  <DrPizza>maybe it only does that for the registry and not pdh
05:45:04  <indutny>nope
05:45:20  <indutny>Counter path should be localized
05:45:27  <indutny>http://support.microsoft.com/?scid=kb%3Ben-us%3B287159&x=11&y=9
05:46:08  <DrPizza>weird
05:46:14  <DrPizza>so even though the registry always has english
05:46:20  <DrPizza>and HKEY_PERFORMANCE_DATA laways has english
05:46:25  <DrPizza>it forces you to use non-english for PDH?
05:46:27  <DrPizza>how stupid
05:46:59  <indutny>HKEY_PERFORMANCE_DATA is localized
05:47:08  <indutny>and yes - it's stupid
05:47:09  <DrPizza>http://msdn.microsoft.com/en-us/library/aa373181%28v=VS.85%29.aspx
05:47:14  <DrPizza>The language identifier is optional. If you do not specify a language identifier, the function returns English strings.
05:47:33  <indutny>if requesting "Counter"
05:47:38  <indutny>but that's just names
05:47:40  <indutny>not values
05:47:47  <indutny>you need to request "2" (i.e System)
05:47:49  <DrPizza>???
05:48:03  <DrPizza>what else gets localized?
05:48:06  <DrPizza>the uptime sure doesn't
05:48:21  <DrPizza>just counter names isn't it?
05:48:30  <indutny>omg, please read docs :)
05:48:47  <DrPizza>i have read the docs and written the code
05:48:51  <DrPizza>I did no thave to deal with all this!
05:50:07  <indutny>of course
05:50:17  <indutny>try switching default language to French
05:50:23  <indutny>your code will be broken AFAIK
05:51:22  <igorzi>erickt: yes, that change needs to be applied to udp as well
05:51:23  <DrPizza>no, I just let it use default english, like it says you can
05:52:53  <indutny>haha :)
05:52:57  <indutny>can you show me your code?
05:53:02  <DrPizza>if I can find it
05:53:07  * isaacsquit (Quit: isaacs)
05:58:55  <indutny>Can "Start Up Time" be zero?
06:00:28  <indutny>looks like yes
06:00:40  <DrPizza>sounds weird
06:02:30  <indutny>"System Up Time"* of course
06:02:36  <indutny>and it's eight zeros
06:02:38  <indutny>wtf
06:05:26  <DrPizza>so you're saying you can't just read "SOFTWARE\\Microsoft\\Windows NT\\CurrentVersion\\Perflib\\009" (on any locale), find the counter IDs named "System" and "System Up Time" and then query HKEY_PERFORMANCE_DATA\system-counter-id and search the returned data for system-uptime-counter-id?
06:05:31  <DrPizza>because that's what the windows source does
06:05:49  <DrPizza>regardless of locale
06:05:51  <DrPizza>009 is hardcoded
06:06:38  <DrPizza>"System" is hardcoded
06:06:43  <DrPizza>"System Up Time" is hardcoded
06:06:50  <DrPizza>all in english
06:07:01  <indutny>nope
06:07:08  <indutny>have you seen that msdn support post?
06:07:17  <DrPizza>have you seen the Windows source code?
06:07:22  <indutny>haha :)
06:07:24  <indutny>no
06:07:29  <DrPizza>SNMP does this
06:07:31  <indutny>are you Microsoft employee?
06:08:22  <DrPizza>indutny: http://www.wired.com/science/discoveries/news/2004/02/62282
06:08:33  <indutny>ah
06:09:23  <DrPizza>if perflib\009 doesn't exist, and doesn't countain english-language counter data then the docs are wrong, and moreover, there's literally no way of knowing which counter has what ID
06:09:46  <indutny>ok write your code
06:09:51  <indutny>and I'll test it on Russian windows
06:09:53  <DrPizza>I will
06:09:54  <DrPizza>on thursday
06:09:54  <indutny>sounds good?
06:10:24  <indutny>DrPizza: can you run "systeminfo" and gist results please?
06:11:40  <DrPizza>https://gist.github.com/1255673
06:13:08  <indutny>wierd
06:13:15  <indutny>aah
06:13:15  <DrPizza>?
06:13:21  <indutny>that was Windows NT only
06:13:30  <DrPizza>what was?
06:14:05  <indutny>ryah: http://en.wikipedia.org/wiki/Uptime#Microsoft_Windows_NT
06:14:08  <indutny>ryah: sorry
06:14:10  <indutny>DrPizza: http://en.wikipedia.org/wiki/Uptime#Microsoft_Windows_NT
06:15:06  <DrPizza>Note: Windows Vista Business 64-bit and Windows 7 do not return a "System Up Time" but "System Boot Time" instead.
06:15:13  <DrPizza>I can't believe it's just vista business
06:15:44  <indutny>aah
06:16:00  <indutny>do you think that's why I can get it?
06:16:03  <DrPizza>probably all vista, win7
06:16:05  <DrPizza>no
06:16:07  <DrPizza>the counter is sitll there
06:19:33  <indutny>Y this stupid value is zero?
06:19:43  <DrPizza>no idea
06:19:44  <DrPizza>sleepy time
06:20:32  <indutny>sleep tight
07:14:08  * felixgequit (Quit: felixge)
07:14:54  * felixgejoined
07:15:07  * felixgequit (Changing host)
07:15:07  * felixgejoined
07:19:19  * felixgequit (Ping timeout: 252 seconds)
08:38:07  * piscisaureusjoined
08:46:48  * piscisaureus_joined
08:46:59  * piscisaureusquit (Ping timeout: 276 seconds)
08:55:54  * piscisaureus_quit (Ping timeout: 255 seconds)
09:09:06  * piscisaureus_joined
09:18:15  * felixgejoined
09:18:31  <felixge>piscisaureus_: https://gist.github.com/21981287f4f7315b1633
09:27:01  * felixgequit (Quit: felixge)
09:32:19  * felixgejoined
09:32:28  * felixgequit (Client Quit)
09:38:32  * piscisaureus_quit (Read error: Connection reset by peer)
09:39:08  * piscisaureus_joined
09:46:55  * piscisaureus_quit (Read error: Connection reset by peer)
09:47:32  * piscisaureus_joined
09:47:38  * piscisaureus_quit (Client Quit)
09:48:26  * felixgejoined
09:48:26  * felixgequit (Changing host)
09:48:26  * felixgejoined
09:49:10  * felixgequit (Client Quit)
11:25:02  <indutny>heya!
11:25:08  <indutny>any joyent guys here?
11:37:16  * felixgejoined
11:37:16  * felixgequit (Changing host)
11:37:16  * felixgejoined
11:39:22  * felixge_joined
11:42:00  * felixgequit (Ping timeout: 252 seconds)
11:55:10  * felixgejoined
11:55:10  * felixgequit (Changing host)
11:55:10  * felixgejoined
11:57:45  * felixge_quit (Ping timeout: 260 seconds)
12:33:51  * felixge_joined
12:33:51  * felixge_quit (Changing host)
12:33:52  * felixge_joined
12:38:01  * felixgequit (Ping timeout: 260 seconds)
12:55:57  * piscisaureusjoined
13:01:33  * felixge_quit (Quit: felixge_)
13:06:21  * piscisaureusquit (Ping timeout: 260 seconds)
13:14:28  * dmkbotjoined
13:15:00  * dmkbotquit (Remote host closed the connection)
13:28:24  * felixgejoined
13:28:51  * felixgequit (Client Quit)
13:49:29  * piscisaureusjoined
14:22:48  <piscisaureus>I just helpen felixge get started on a more reasonable spawn_sync implementation (that's built on top of multiplicity)
14:22:52  <piscisaureus>*helped
14:22:56  <piscisaureus>https://github.com/joyent/libuv/tree/spawn_sync
14:23:32  <piscisaureus>But then I realized that libev only supports child watchers on the main loop, so it's prolly not going to work on unix
14:23:47  <piscisaureus>is there a way around this that you guys (ryah bnoordhuis) can think of?
14:25:08  <piscisaureus>Talk here - I'll be observing the logs
14:30:01  * piscisaureusquit (Ping timeout: 260 seconds)
14:44:51  * erickt_joined
15:16:54  <indutny>how can I build libuv on macos?
15:24:44  <erickt_>indutny: I tend to run "gyp_uv -f make && cd out && make"
15:24:50  <indutny>ah
15:24:52  <indutny>cool
15:24:56  <indutny>is it working?
15:25:03  <indutny>gyp_uv -f make && make wasn't
15:25:03  <erickt_>works for me :)
15:25:07  <indutny>ok, great
15:25:09  <indutny>thanks
15:25:17  <erickt_>did you check out gyp into build/gyp?
15:27:49  <indutny>erickt_: yes
15:29:40  * erickt_quit (Quit: erickt_)
15:31:09  <indutny>erickt: yt?
16:01:24  * felixgejoined
16:01:24  * felixgequit (Changing host)
16:01:24  * felixgejoined
16:02:32  * felixgequit (Client Quit)
16:12:51  * dmkbotjoined
16:54:28  * dmkbotquit (Remote host closed the connection)
17:24:39  * dmkbotjoined
18:02:51  <indutny>anyone on freebsd? or on openbsd here?
18:14:36  <dmkbot>joyent/node: indutny: Debugger docs: round 1 - https://github.com/joyent/node/issues/1809
20:01:03  * dmkbotquit (Remote host closed the connection)
20:01:27  * dmkbotjoined
20:10:38  * brsonjoined
20:22:21  * bnoordhuisjoined
22:08:04  * mattstevensjoined
22:44:16  <CIA-53>libuv: Ben Noordhuis master * r914a6fa / src/unix/sunos.c : sunos: pull in uv__set_sys_error prototype - http://git.io/10m1YQ