00:05:51  * mralephquit (Quit: Leaving.)
01:35:33  <CIA-53>node: Ben Noordhuis master * r6ee676c / src/platform_cygwin.cc : cygwin: fix return of uninitialized variable - http://git.io/YAwujg
01:35:33  <CIA-53>node: Ben Noordhuis master * rd91f64f / src/platform_cygwin.cc : cygwin: fix new[]/delete mismatch - http://git.io/92AEIw
01:36:51  <CIA-53>libuv: Ben Noordhuis master * r01441ab / src/unix/udp.c : unix: fix close() of potentially uninitialized fd - http://git.io/8QUXoA
01:36:51  <CIA-53>libuv: Ben Noordhuis master * rd31a57d / src/unix/cares.c : unix: fix NULL pointer dereference - http://git.io/WgdXzQ
01:53:14  * brsonquit (Quit: leaving)
01:53:34  * ericktjoined
01:57:35  <piscisaureus>bnoordhuis: don't fix cygwin on master ;-)
01:57:42  <piscisaureus>bnoordhuis: fix it on 0.4 - fine
01:57:56  <bnoordhuis>oh, this is nothing serious
01:58:08  <bnoordhuis>that is, potentially serious bug but not a big change
01:58:23  <piscisaureus>bnoordhuis: this thing is, cygwin is dying
01:58:28  <piscisaureus>actually it is already dead
01:58:35  <bnoordhuis>that's what they said about netbsd
01:58:43  <piscisaureus>except for on the 0.4 branch
01:58:45  <bnoordhuis>and look what's running on my toaster now
01:58:55  <piscisaureus>bnoordhuis: when I have time I'm gonna remove all cygwin crap :-)
01:58:57  <piscisaureus>:p
01:59:20  <bnoordhuis>good point about the 0.4 branch, i should back-port that shizzle
01:59:28  <piscisaureus>yes, that makes sense
02:00:30  <CIA-53>node: Ben Noordhuis v0.4 * raff91f7 / src/platform_cygwin.cc : cygwin: fix return of uninitialized variable - http://git.io/HFIvtw
02:00:31  <CIA-53>node: Ben Noordhuis v0.4 * r942b92a / src/platform_cygwin.cc : cygwin: fix new[]/delete mismatch - http://git.io/Jjx45Q
02:01:56  <bnoordhuis>btw, http://tinyurl.com/GitSurvey2011
02:11:26  <bnoordhuis>okay, off to bed
02:11:29  <bnoordhuis>sleep tight, piscisaureus
02:11:52  <piscisaureus>ah
02:11:57  <piscisaureus>so I'm all alone
02:12:01  <piscisaureus>goodnight bnoordhuis
02:12:20  <CIA-53>libuv: Bert Belder master * r985f1a1 / include/uv-private/uv-win.h : cleanup - http://git.io/8SQi9w
02:12:20  <CIA-53>libuv: Bert Belder master * r77bc00e / src/win/pipe.c : win: uv_pipe_queue_read should set read_pending flag - http://git.io/utMQPA
02:12:56  <CIA-53>libuv: Bert Belder wintty * r8c1f528 / (5 files in 2 dirs): win: TTY work - WIP - http://git.io/H8BEww
02:13:20  <piscisaureus>bnoordhuis: I want to see the git survey results of last year
02:13:28  <piscisaureus>instead of filling it out *again*
02:17:10  * bnoordhuisquit (Ping timeout: 276 seconds)
02:29:59  <erickt>Hi #libuv. So I was working on sockets for rust, and I realized that libuv can't really leverage using getaddrinfo to automatically pick a ipv4/ipv6/ipv6-to-ipv4
02:31:20  <erickt>What do you all think about renaming uv_bind to uv_bind4, and allowing uv_bind to take an abstract sockaddr?
02:58:57  <piscisaureus>hmm. I think we could could get rid of passing sockaddr's by value at all.
03:00:28  <piscisaureus>however I'm not going to decide on my own now.
03:01:10  <piscisaureus>I'd suggest to create an issue and bug people to comment.
04:21:01  <indutny>bnoordhuis: hi
04:21:06  <indutny>nice tty/win commits
04:21:27  <indutny>actually, nice to see that you're storing handles not fds
04:51:27  * piscisaureusquit (Ping timeout: 248 seconds)
09:53:29  <indutny>anyone around?
10:45:47  * mralephjoined
13:06:29  * mikealjoined
13:25:47  * mikealquit (Quit: Leaving.)
14:40:26  * piscisaureusjoined
16:19:45  <piscisaureus>ryah: what tty modes did we define?
16:20:08  <piscisaureus>afaict unix/tty.c always sets the same mode?
16:47:36  <indutny>piscisaureus: yeah
16:47:41  <indutny>piscisaureus: actually there're two modes
16:47:45  <indutny>raw and not-raw :P
16:47:54  <indutny>piscisaureus: yt?
16:48:02  <piscisaureus>indutny: yes
16:48:11  <indutny>I was digging this code out
16:48:19  <indutny>https://github.com/joyent/node/blob/master/src/node_stdio.cc#L54
16:48:42  <piscisaureus>indutny: we decided that there should be more modes in libuv - but it doesn't seem that the unix tty has them yet
16:48:43  <indutny>that's how it used in node
16:48:57  <indutny>piscisaureus: yep, it's ignoring that argument
16:49:20  <indutny>piscisaureus: ok, though. I don't know any other modes
16:49:32  * piscisaureustries to remember
16:50:00  <piscisaureus>I think for the time being I will just make uv_tty_set_mode set the raw mode like the legacy bindings do
16:50:13  <piscisaureus>add moar modes later
16:50:26  <indutny>heh
17:00:22  <piscisaureus>__builtin_clzl <-- is that generally supported by gcc?
17:01:52  <mraleph>should be
17:12:00  <indutny>http://gcc.gnu.org/onlinedocs/gcc-4.4.2/gcc/Other-Builtins.html
17:14:32  <piscisaureus>Oh nice, no harm in using that then :-) thanks
18:07:23  <indutny>ryah: yt?
18:17:34  * isaacsjoined
18:57:22  <ryah>piscisaureus: just do two modes
18:57:41  <piscisaureus>ok
19:01:42  <ryah>indutny: still having problems with "bt"
19:02:06  <ryah>indutny: is it possible when the debugger attaches to a remote process that it doesn't stop it?
19:02:49  <ryah>indutny: the windows signal we need to handle at the libuv layer
19:06:05  <ryah>indutny: i'll land the whole lot for now...
19:07:45  <CIA-53>node: Fedor Indutny master * r9e09fc0 / (lib/_debugger.js src/node.cc):
19:07:46  <CIA-53>node: more cli options
19:07:46  <CIA-53>node: * node debug localhost:5858 - connects to remote debugger
19:07:46  <CIA-53>node: * node debug -p `pgrep node` - connects to running process
19:07:46  <CIA-53>node: * Fixed double-run of debugger on SIGUSR1 - http://git.io/8kUEQA
19:07:47  <CIA-53>node: Fedor Indutny master * r2c9bcb2 / src/node.cc : move debugger_running to correct place - http://git.io/BY51wQ
19:07:47  <CIA-53>node: Fedor Indutny master * r9b6acc2 / lib/_debugger.js : handle backtrace errors - http://git.io/uMyDBA
19:07:47  <CIA-53>node: Fedor Indutny master * r26aab0d / src/node.cc :
19:07:47  <CIA-53>node: EnableDebug immediatly on SIGUSR1
19:07:48  <CIA-53>node: Don't wait for script to break somewhere, because that may not happen if
19:07:48  <CIA-53>node: execution is inside event-loop, not in v8.
19:08:11  <indutny>heh
19:08:42  <indutny>ryah: it doesn't stop it if it's inside event-loop
19:08:42  <CIA-53>node: Add '\n' to the end of 'debugger listening...' message - http://git.io/XYDbeA
19:08:42  <CIA-53>node: Fedor Indutny master * r82d0ac7 / src/node.cc : Ctrl+Break on windows starts debugger server - http://git.io/v5cvww
19:08:46  <indutny>not in javascript land
19:09:11  <indutny>for example http_simple.js
19:09:33  <indutny>when it's waiting for http request - no javascript is being executed
19:09:41  <indutny>so debugger can't break anywhere
19:09:52  <ryah>indutny: but good work - this is really useful
19:10:08  <indutny>thanks
19:10:18  <indutny>ryah: btw, no hope for windows' remote process debugging
19:10:20  <indutny>no IPC
19:10:21  <indutny>:D
19:10:30  <indutny>only emulation
19:10:38  <indutny>and that won't work for services
19:10:40  <ryah>indutny: it would be nice if you prefixed your first line with "debugger: "
19:10:45  <ryah>in the commit message
19:10:48  <indutny>ryah: ok
19:10:58  <indutny>in next commits
19:11:03  <ryah>indutny: k ;)
19:11:10  <indutny>so lets see that bt stuff...
19:11:29  <ryah>indutny: https://github.com/joyent/node/pull/1766#issuecomment-2192430
19:11:50  <indutny>ryah: is that run with http_simple.js ?
19:11:54  <ryah>indutny: yes
19:12:32  <ryah>indutny: we can send a Ctrl+break in windows remotely
19:12:42  <indutny>ryah: realy?
19:12:47  <indutny>s/realy/really
19:13:03  <indutny>aah
19:13:12  <ryah>indutny: http://msdn.microsoft.com/en-us/library/windows/desktop/ms683155(v=vs.85).aspx
19:13:26  <indutny>ryah: http://msdn.microsoft.com/en-us/library/ms683155.aspx
19:13:31  <indutny>yeah
19:13:37  <indutny>I forgot about this
19:13:40  <indutny>thanks for reminding
19:14:02  <ryah>indutny: but we need to discuss with piscisaureus and igorzi about if this is what we want...
19:14:07  <ryah>i kind of think we do
19:14:14  <indutny>k
19:14:20  <indutny>ryah: I looked over other variants
19:14:25  <ryah>i think we want to have a unix signal emulation layer in windows...
19:14:27  <indutny>ryah: named pipes and etc
19:14:31  <indutny>ryah: yeah
19:14:52  <piscisaureus>indutny: the other alternative is most likely a mailslot :-p
19:15:11  <indutny>piscisaureus: but that'll require us to run thread
19:15:15  <ryah>especially now that we have a vt100 emulation layer :)
19:15:29  <piscisaureus>ryah: I am still coding the ansi escape code parsing
19:15:31  <piscisaureus>going well tho
19:15:35  <ryah>piscisaureus: nice
19:15:47  <indutny>piscisaureus: anyway, I can't understand difference between mailslot and named pipes
19:15:49  <piscisaureus>CTRL+BREAK is probably nice - reminds me of good old QB45 days
19:15:55  <indutny>they looks too similar to each other
19:16:03  <ryah>piscisaureus: feel free to add a test program to libuv even if it can't be checke automatiatically
19:16:08  <piscisaureus>indutny: mailslots are like datagram sockets - they're not connection oriented
19:16:10  <ryah>piscisaureus: at least we can share screenshots of the output for now
19:16:18  <indutny>piscisaureus: aah
19:16:55  <indutny>ryah: have you received email notification from Brendan Rampersad? :D
19:17:02  <indutny>hahaa
19:17:09  <ryah>yes
19:17:12  <ryah>who is that?
19:17:54  <ryah>piscisaureus: are we agreed on uv_write2 and uv_read2_start for ipc?
19:18:04  <ryah>piscisaureus: instead of the ipc stream?
19:18:16  <piscisaureus>ryah: I guess
19:18:19  <ryah>piscisaureus: i've already started - so i hope :)
19:18:20  <indutny>ryah: I dunno
19:18:40  <ryah>piscisaureus: we can think of better names later
19:18:55  <ryah>piscisaureus: but uv_write' and uv_read_start'
19:18:57  <ryah>(prime)
19:19:21  <piscisaureus>ryah: I haven't really thought about the consequences for the windows implementation. I guess it should be doable - the only problem is where to store all this extra state that an overlay protocol requires.
19:19:21  <piscisaureus>But that's probably a minor issue.
19:19:43  <ryah>(not sure if this prime notation is only known to math people or is known more generally)
19:21:01  <piscisaureus>In my world ' usually means the derivate of some kind.
19:21:01  <piscisaureus>e.g. v'= fn(v)
19:21:58  <indutny>ryah: https://github.com/indutny/node/tree/fix-debugger-bt
19:22:15  <indutny>https://github.com/joyent/node/pull/1768
19:22:17  <indutny>actually
19:22:59  <indutny>piscisaureus: ryah: will you guys work on IPC messaging stuff?
19:23:16  <indutny>or should I hard code GenerateConsoleCtrlEvent somewhere in node.cc
19:24:04  <piscisaureus>"somewhere" should be in process.kill and child_process.kill
19:24:21  <piscisaureus>e.g. generate this event when the uisers kills with SIGBRK or something
19:25:34  <ryah>yes, i agree with piscisaureus
19:25:38  <indutny>yeah
19:25:50  <indutny>hard-coding ftw!
19:25:51  <ryah>indutny: it might be worthwhile to just add uv_kill now
19:25:51  <indutny>:)
19:26:05  <indutny>ryah: heh, you decide - me implements
19:26:20  <indutny>:)
19:26:31  <ryah>i have one question about GenerateConsoleCtrlEvent - the second argument is dwProcessGroupId
19:26:44  <ryah>is putting a PID in there going to work/
19:27:21  <indutny>ryah: probably no
19:27:32  <indutny>ryah: I don't understand how that should work in windows security model
19:27:42  <indutny>ryah: is it ok to send Ctrl+Break from any process to any process
19:27:43  <indutny>?
19:27:50  <indutny>piscisaureus: what do you think?
19:29:08  <ryah>indutny: you can just try it :)
19:29:16  <indutny>haha
19:29:19  <indutny>that's not so simple
19:29:22  <piscisaureus>indutny: what are you talking about? windows security mode???? :-)
19:29:23  <indutny>I need to run vmware
19:29:28  <indutny>piscisaureus: model
19:29:58  <piscisaureus>indutny: but yeah it'll work, especially when the other process uses the same console.
19:34:17  <indutny>gtg, going to sleep... need to wake up in 5 hours
19:34:20  <indutny>thanks everyone
19:34:23  <indutny>see you tomorrow!
19:35:39  <ryah>indutny: night
19:38:38  <CIA-53>node: Fedor Indutny master * r02e0a0a / doc/api/debugger.markdown :
19:38:38  <CIA-53>node: debugger docs
19:38:38  <CIA-53>node: Fixes #1767 - http://git.io/QBBO7A
19:48:03  * ircretaryjoined
19:49:28  * isaacsquit (Quit: isaacs)
20:42:00  <ryah>people are pretty stupid: http://www.reddit.com/r/programming/comments/kq27q/issue_847_v8_1gb_memory_limit_v8_javascript_engine/c2mbfad
20:42:06  <piscisaureus>vt100 colors work. now cursor movement
20:42:10  <ryah>http://www.reddit.com/r/programming/comments/kq27q/issue_847_v8_1gb_memory_limit_v8_javascript_engine/c2mfccg
20:42:43  <ryah>im not complaining that they don't understand the problem - but more that they seem convinced that they do understand the problem :)
20:43:03  <piscisaureus>yes.
20:44:50  <piscisaureus>Go hit him with a stick
20:49:01  <ryah>piscisaureus: you want to land your branch?
20:49:24  <piscisaureus>ryah: hmm. it has loose ends
20:49:28  <piscisaureus>you want me to?
20:49:33  <piscisaureus>I mean, it compiles :-)
20:49:34  <ryah>what are the loose ends?
20:49:40  <mraleph>frankly speaking using signed types in C++ is pita. I dig size_t but I don't dig idiotic warnings about comparing signed to unsigned.
20:50:11  <piscisaureus>ryah: parts of the vt100 parser. uv_tty_set_mode not implemented. close/shutdown not implemented. all reads are raw.
20:50:40  <piscisaureus>ryah: let me push
20:50:49  <ryah>oh - well maybe you should do some more work then
20:51:40  <ryah>i'm eager to land new-tty-binding branch in node
20:52:18  <CIA-53>libuv: Bert Belder wintty * r431daee / (5 files in 2 dirs): win: TTY work - WIP - http://git.io/0uj7LQ
20:52:38  <piscisaureus>ryah: I'm approaching the end quickly. set_mode is easy
20:52:49  <piscisaureus>Just need to get cursor movement in place
20:52:53  <ryah>piscisaureus: k
20:53:12  <ryah>it's going to be really nice to have this
20:54:39  <piscisaureus>I'm also going to need a strategy to break this up -> https://github.com/joyent/libuv/blob/431daee65f080730d553b39de5f65adf9da72a83/src/win/tty.c#L590-850
21:14:08  <DrPizza>piscisaureus: is that true that the windows console doesn't like surrogate pairs, or is it just that most fonts don't support them?
21:14:41  <piscisaureus>no I put that there for fun :-(
21:15:09  <piscisaureus>if you write a surrogate pair you get 2 replacement characters
21:15:13  <DrPizza>huh
21:15:16  <piscisaureus>so it doesn't support it
21:15:20  <DrPizza>that sounds kinda broken
21:18:20  * isaacsjoined
21:29:55  * ericktquit (Quit: erickt)
21:47:49  * brsonjoined
22:24:59  * isaacsquit (Ping timeout: 256 seconds)
22:32:27  * ericktjoined
22:36:45  <erickt>Does Windows not automatically bind a socket inside ConnectEx? I see uv_tcp_connect does a bind if it hasn't been done already
22:46:13  <piscisaureus>erickt:
22:46:13  <piscisaureus>http://msdn.microsoft.com/en-us/library/windows/desktop/ms737606%28v=vs.85%29.aspx ->
22:46:13  <piscisaureus>"s [in] A descriptor that identifies an unconnected, previously bound socket. See Remarks for more information."
22:46:32  <erickt>ah, thanks
22:50:26  <erickt>I'm poking around at making bind/connect accept ipv4 and ipv6 sockaddrs, and I found that the windows code needs a nice code deduplication pass :)
22:57:05  <CIA-53>libuv: Bert Belder wintty * r46915bd / (5 files in 2 dirs): win: TTY work - WIP - http://git.io/huXuyQ
22:57:26  <piscisaureus>^-- almost there. Anyone feels like reviewing 1100 LOC?
23:06:02  * mralephquit (Quit: Leaving.)
23:07:36  <erickt>Would anyone mind if I rename any private windows functions to uv__foo, like how the unix functions are done?
23:13:32  <piscisaureus>I wouldn''t
23:26:11  * brsonquit (Ping timeout: 256 seconds)
23:27:31  * brsonjoined
23:27:57  <piscisaureus>ryah: here?