00:00:04  * loladiroquit (Quit: loladiro)
00:02:36  * bentkusquit (Read error: No route to host)
00:03:21  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
00:23:46  * joshthecoderjoined
00:30:06  * TooTallNatequit (Quit: Computer has gone to sleep.)
00:32:12  * TooTallNatejoined
00:34:12  * mikealquit (Quit: Leaving.)
00:37:01  * joshthecoderquit (Quit: Leaving...)
01:02:15  * TooTallNatequit (Quit: Computer has gone to sleep.)
01:10:05  * TooTallNatejoined
01:31:47  * loladirojoined
02:03:47  * bnoordhuisquit (Ping timeout: 244 seconds)
02:21:11  * brsonquit (Quit: leaving)
02:44:50  * TooTallNatequit (Quit: Computer has gone to sleep.)
02:59:25  * mcavagejoined
03:00:37  * mmaleckiquit (Ping timeout: 244 seconds)
03:16:37  * brsonjoined
03:22:43  * AvianFluquit (Quit: AvianFlu)
03:32:50  * TooTallNatejoined
03:43:20  * TooTallNatequit (Quit: Computer has gone to sleep.)
03:45:41  * TooTallNatejoined
03:59:05  * chobi_echanged nick to chobi_e_
04:04:31  * avalanche123joined
04:14:25  * mcavagequit (Remote host closed the connection)
04:24:55  * philips_quit (Excess Flood)
04:26:14  * philips_joined
04:49:38  * mbalhoquit (Ping timeout: 244 seconds)
04:50:09  * paddybyersjoined
04:52:02  * paddybyersquit (Client Quit)
05:06:31  * avalanche123quit (Quit: Computer has gone to sleep.)
05:57:43  * TooTallNatequit (Quit: ["Textual IRC Client: www.textualapp.com"])
06:18:55  * brsonquit (Remote host closed the connection)
06:27:13  * chobi_e_changed nick to chobi_e
06:28:00  * loladiroquit (Quit: loladiro)
06:34:40  * avalanche123joined
06:47:51  * avalanche123quit (Quit: Computer has gone to sleep.)
06:48:57  * philips_quit (*.net *.split)
06:48:58  * Benviequit (*.net *.split)
06:48:59  * CIA-131quit (*.net *.split)
06:48:59  * dscapequit (*.net *.split)
06:48:59  * mitsuhikoquit (*.net *.split)
07:06:07  * chobi_echanged nick to chobi_e_
07:11:57  * Benviejoined
07:12:43  * philips_joined
07:12:52  * CIA-131joined
07:12:52  * dscapejoined
07:12:52  * mitsuhikojoined
07:15:35  * avalanche123joined
07:29:51  * avalanche123quit (Quit: Computer has gone to sleep.)
07:48:35  * rendarjoined
09:01:18  * TheJHjoined
09:10:00  * chobi_e_changed nick to chobi_e
09:16:49  * chobi_echanged nick to chobi_e_
09:22:46  * chobi_e_changed nick to chobi_e
09:22:58  * FUUFREENODESTOPBjoined
09:23:09  * FUUFREENODESTOPBchanged nick to `3rdEden
09:23:29  * `3rdEdenquit (Client Quit)
09:46:37  * chobi_echanged nick to chobi_e_
10:01:16  * chobi_e_changed nick to chobi_e
10:09:43  * TheJHchanged nick to Guest38351
10:15:27  * hzjoined
10:25:59  * mmaleckijoined
10:50:23  * stagasjoined
12:13:17  * AutowiredConstrujoined
12:24:59  * piscisaureus_joined
12:25:11  <piscisaureus_>\o
12:25:29  <piscisaureus_>hey indutny, did you figure out what the issue was with the hanging stat requests?
12:25:49  <indutny>piscisaureus_: hoya
12:25:51  <indutny>yes
12:25:55  <indutny>it was bug in libfakeroot
12:25:58  <piscisaureus_>indutny: nice. what was it
12:26:02  <piscisaureus_>libfakeroot?
12:26:40  <indutny>yeah
12:26:47  <piscisaureus_>well, good
12:26:52  <indutny>those guys were running node under fakeroot
12:27:04  <indutny>and it has deadlock in stat() call
12:27:13  <piscisaureus_>oh, haha
12:27:19  <indutny>yeah, definitely haha
12:27:24  <indutny>piscisaureus_: https://github.com/indutny/tlsnappy
12:27:30  <indutny>it's working ^
12:27:31  <indutny>though, leaking memory a little
12:27:36  <indutny>and doing a lot of allocations
12:27:39  <indutny>but it's working :)
12:27:45  <indutny>800 req/sec in one node process
12:27:48  <indutny>using 4 threads
12:28:17  <piscisaureus_>how much does it do with 1 thread?
12:28:58  <indutny>one sec
12:29:00  <indutny>I'll test
12:29:34  <indutny>385
12:30:00  * Guest38351quit (Quit: goodbye)
12:30:13  <indutny>the same node build
12:30:16  <indutny>and https module
12:30:19  <indutny>331 req/sec
12:30:44  <piscisaureus_>right, ok, so it could scale somewhat better
12:30:48  <piscisaureus_>indutny: but, nice :-)
12:30:51  <piscisaureus_>indutny: good work
12:30:51  <indutny>haha
12:30:54  <indutny>yeah, it can scale
12:31:02  <indutny>I'm allocating 4 ring buffers on each connection
12:31:08  <indutny>which is definitely wasting time
12:31:09  <piscisaureus_>indutny: do you want to make this part of node at some point?
12:31:13  <indutny>piscisaureus_: dunno
12:31:23  <indutny>piscisaureus_: I would like to remove 'tls' in node at some point
12:31:33  <indutny>piscisaureus_: separate upgradable parts are much better
12:32:14  <indutny>and also
12:32:20  <indutny>it doesn't support SNI and NPN for now
12:32:34  <piscisaureus_>indutny: true, but binary addons also means more headache
12:32:35  <indutny>which should not be a problem, but just isn't implemented yet
12:32:44  <piscisaureus_>indutny: right
12:32:46  <indutny>piscisaureus_: yeah
12:32:49  <piscisaureus_>let's keep it separate then
12:32:50  <indutny>agreed
12:32:58  <indutny>let's finish it :)
12:33:00  <indutny>and fix memory leaks
12:33:09  <piscisaureus_>let's make it work on windows :-p
12:33:13  <indutny>sockets are not getting released
12:33:17  <indutny>piscisaureus_: it should work :)
12:33:22  <indutny>piscisaureus_: it's using only uv's APIs
12:33:29  <indutny>nothing unix specific AFAIK
12:33:32  <piscisaureus_>well, it complains about opensslconf.h not being found
12:33:43  <piscisaureus_>indutny: on windows openssl doesn't get exported atm
12:33:46  <piscisaureus_>which is PITA
12:33:50  <piscisaureus_>i should go fix that
12:34:16  <indutny>oh
12:34:26  <indutny>yeah, that's mostly your fault :D
12:34:37  <piscisaureus_>yep
12:34:41  <indutny>piscisaureus_: I would be really pleased if you'll help me to figure out why it leaks memory
12:34:54  <piscisaureus_>indutny: do you know what leaked?
12:35:04  <piscisaureus_>you use valgrind i take it>
12:35:41  <indutny>it's not so simple :)
12:36:00  <piscisaureus_>indutny: why not?
12:36:14  <indutny>piscisaureus_: idk
12:36:22  <indutny>last time I check it hasn't showed me anything useful
12:36:28  <indutny>i.e. only node and v8 leaks
12:36:34  <piscisaureus_>oh like that
12:36:42  <indutny>and about 5 mb of leaked memory
12:36:44  <piscisaureus_>I'll start by reading the code
12:36:52  <indutny>while I see rss around 900 mb
12:36:56  <indutny>after 500000 connections
12:37:09  <piscisaureus_>indutny: was that v8 heap?
12:37:22  <indutny>piscisaureus_: I presume not
12:37:31  <piscisaureus_>buffers?
12:37:48  <indutny>lets see
12:37:51  <indutny>running valgrind again
12:38:08  <indutny>oh, much better
12:38:08  <piscisaureus_>it could be that somehow you're retaining references to buffers and/or slabs
12:38:10  <indutny>40mb in use at exit
12:38:37  <piscisaureus_>if you use master you can use node-heapdump
12:38:45  <piscisaureus_>and inspect the dump with chrome
12:39:01  <indutny>nope, I'm not on master
12:39:08  <indutny>it's hard to make binary addons working with it
12:39:22  <piscisaureus_>well, it's just the v8 version that matters really
12:39:40  <piscisaureus_>node-heapdump works with 3.11 but it uses an older snapshot format that chrome can't read
12:40:26  <indutny>ah
12:40:28  <indutny>ok
12:40:38  <piscisaureus_>it would be nice if we had a BIO implementation that used a buffer as backing store
12:40:44  <piscisaureus_>so we could get rid of the unnecessary copy
12:41:42  <piscisaureus_>speaking of garbage collection
12:41:46  * piscisaureus_booted AutowiredConstru (See You - Kick sponsored by www.trillian.im)
12:41:51  <indutny>piscisaureus_: yay
12:41:53  <indutny>finally!
12:41:59  <piscisaureus_>shit
12:42:03  <piscisaureus_>I forgot to capture his IP
12:42:06  <indutny>haha
12:42:16  <indutny>who is he?
12:42:28  <indutny>your alter ego?
12:44:20  <piscisaureus_>I don't know
12:44:20  <piscisaureus_>I don't know
12:44:34  <piscisaureus_>Not my alter ego
12:45:40  <indutny>ok
12:45:56  <indutny>and btw
12:46:03  <indutny>those benchmark results are w/o openssl patches
12:46:07  <indutny>i.e. w/o assembly
12:46:12  <piscisaureus_>yeah
12:46:20  <piscisaureus_>I have to figure this slow aes issue too
12:46:23  <piscisaureus_>but I am clueless
12:46:30  <piscisaureus_>I probably need to mail the openssl list
12:46:35  <piscisaureus_>it could just be a canard
12:47:19  <indutny>meh
12:47:21  <indutny>nothing interesting
12:47:34  <indutny>in valgrind logs
12:47:48  <indutny>I don't think it leaks much on low amount of requests
12:47:56  <piscisaureus_>did you run with --leak-check=full ?
12:48:13  <indutny>yes
12:48:21  <indutny>there're one big buffer
12:48:26  <indutny>which looks like a tls buffer
12:48:33  <indutny>and nothing interesting
12:48:40  <piscisaureus_>ah, righ
12:48:43  <piscisaureus_>well, I'll take a look
12:48:54  <piscisaureus_>fixing the build first
12:48:55  <indutny>thanks
12:49:38  <indutny>piscisaureus_: btw, pull latest update just in case
12:51:12  <piscisaureus_>k
12:51:42  <piscisaureus_>mode #libuv +b *!*@84.60.117.70
12:53:01  <deoxxa>what's with the ban?
12:53:08  <deoxxa>is this guy a jerk? :<
12:53:29  <piscisaureus_>deoxxa: he's annoying
12:53:32  <piscisaureus_>#libuv is a private party
12:54:45  <piscisaureus_>or she (hey, not being sexist here)
12:54:56  * deoxxanods
12:54:58  <deoxxa>girls can be jerks too
13:16:37  * TheJHjoined
13:20:58  <mraleph>indutny: piscisaureus_: pro-tip (might be unrelated but…) if you see a strange leak that might even remotely be caused by JavaScript try running with optimizations disabled (—no-crankshaft).
13:26:15  <piscisaureus_>mraleph: I have never found a leak that was definitely caused by v8
13:26:22  <piscisaureus_>mraleph: but it's good to know - thank
13:26:25  <piscisaureus_>s
13:27:32  * TheJHquit (Quit: goodbye)
13:30:06  * TheJHjoined
13:30:58  * kristatejoined
13:53:54  * TheJHquit (Remote host closed the connection)
13:57:31  * TheJHjoined
14:02:39  <indutny>mraleph: thanks for protip
14:02:59  <indutny>mraleph: I think it's not related to js
14:03:06  <mraleph>good
14:07:51  <piscisaureus_>why is libuv so popular in japan?
14:08:20  <mmalecki>mraleph: hey, what hotel are you staying in?
14:08:27  <indutny>piscisaureus_: dunno
14:08:30  <indutny>why do you think it is?
14:08:45  <mraleph>mmalecki: the one that was on the #lxjs page.
14:08:51  <mraleph>mmalecki: Gus something
14:09:03  <mraleph>Gat Rossio
14:09:08  <mmalecki>mraleph: ah, the same one I'm staying in. also, nobody remembers the full name, ever.
14:09:23  <deoxxa>piscisaureus_: lol what (...wwwnanda)
14:10:01  <mmalecki>mraleph: I'm grabbing a beer on 27th with few folks I know, I'll be sure to hit you up on twitter :)
14:10:07  <piscisaureus_>indutny: that's why
14:10:11  <piscisaureus_>https://twitter.com/search/realtime/?q=%2Blibuv&src=typd
14:10:20  <mraleph>mmalecki: sure!
14:11:07  <mraleph>if I manage to survive until 27th that is :-)
14:11:28  <mmalecki>haha, I hope you will
14:19:23  * mmaleckiquit (Quit: leaving)
14:27:37  * chobi_echanged nick to chobi_e_
14:27:54  <indutny>piscisaureus_: ok
14:28:06  <indutny>piscisaureus_: so have you managed to build it on win32?
14:28:13  <piscisaureus_>yes
14:28:25  <piscisaureus_>hacks and all that
14:28:50  <piscisaureus_>must ... go .. fix .. node
14:29:48  * loladirojoined
14:30:41  <indutny>piscisaureus_: ok
14:30:44  <indutny>piscisaureus_: :)
14:31:01  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
14:31:41  * piscisaureus_joined
14:32:28  <piscisaureus_>bbl
14:32:30  * piscisaureus_quit (Client Quit)
14:46:32  * mcavagejoined
15:01:38  * loladiroquit (Quit: loladiro)
15:11:00  * philips_quit (Excess Flood)
15:11:46  * philips_joined
15:24:13  * AvianFlujoined
15:25:57  * mcavagequit (Remote host closed the connection)
16:13:23  * hzquit (Ping timeout: 272 seconds)
16:35:44  * mmaleckijoined
17:20:16  * mmaleckiquit (Ping timeout: 264 seconds)
17:48:54  <ryah>indutny: does your fs events stuff work?
17:49:12  <ryah>does it poll the CF loop in another thread?
17:49:15  <ryah>how does it work?
17:50:51  <indutny>ryah: yes
17:50:52  <tjfontaine>ya that's what it seems to do
17:50:54  <indutny>ryah: it polls CF
17:51:09  <indutny>ryah: though I've implemented a way to poll it with kqhe
17:51:11  <indutny>kqueue
17:51:15  <indutny>but it's too hacky for libuv
17:51:51  <ryah>indutny: nice - is it going into v0.8 ?
17:52:28  <ryah>is it much faster than the kqueue-based fs notifcation?
17:52:38  <ryah>what prompted you to do this?
17:53:07  <ryah>is it just that you want recursive notifications?
17:56:03  * avalanche123joined
18:22:09  <indutny>ryah: no it's not going into 0.8
18:22:13  <indutny>ryah: because it's an ABI change
18:22:28  <indutny>ryah: it's not faster, just handling directory changes better
18:22:36  <indutny>i.e. it supports changes in nested directory tree
18:22:42  <indutny>recursive notifications
18:22:53  <indutny>and on OSX >= 10.7 it emits filesnames
18:22:57  <indutny>filenames*
18:34:14  * hzjoined
18:37:31  * TooTallNatejoined
19:01:46  * loladirojoined
19:03:05  * paddybyersjoined
19:03:10  * loladiroquit (Client Quit)
19:13:54  * avalanche123quit (Quit: Computer has gone to sleep.)
19:24:06  * loladirojoined
19:27:21  * paddybyersquit (Quit: paddybyers)
19:30:59  * kristatequit (Ping timeout: 272 seconds)
19:37:43  * kristatejoined
19:41:21  * paddybyersjoined
19:41:41  * bnoordhuisjoined
19:47:50  * mikealjoined
19:59:13  * paddybyersquit (Quit: paddybyers)
20:02:16  * piscisaureus_joined
20:02:21  <piscisaureus_>\o
20:02:39  <piscisaureus_>indutny: does fsevent really catch recursive changes now>
20:02:52  <piscisaureus_>indutny: because that's not really supposed to happen ...
20:03:18  <piscisaureus_>indutny: i just wanted this so it would actually work (kqueue often does not catch file changes when you're watching a dir)
20:07:59  * paddybyersjoined
20:09:55  * paddybyersquit (Client Quit)
20:25:31  <saghul>piscisaureus_ indutny FWIW, fs_event_watch_file_current_dir fails for me on Snow Leopard every now and then
20:25:58  <piscisaureus_>saghul: with fsevents? or befor?
20:26:08  <saghul>with fsevents
20:26:41  <saghul>I don't remember if it did before, but I can revert the fsevents and check it out
20:28:46  <piscisaureus_>saghul: sure
20:29:01  <piscisaureus_>saghul: well maybe you can tell me how it fails first?
20:29:36  <saghul>piscisaureus_ oh, right :-) it faisl with timeout
20:30:12  <piscisaureus_>ah, so probably it just "hangs" while waiting for an event that never comes
20:30:20  <piscisaureus_>probably a race condition
20:31:46  <saghul>piscisaureus_ yep, looks like it
20:31:53  <piscisaureus_>ya
20:32:08  <piscisaureus_>it would be nice if it wasn't racy but indutny dismissed my fears :-)
20:34:10  <indutny>piscisaureus_: haha
20:34:15  <indutny>piscisaureus_: racy racy
20:34:26  <saghul>hum, I just reverted the fsevents stuff and still fails sometimes
20:34:31  <indutny>haha
20:34:33  <indutny>racy racy
20:34:35  <indutny>piscisaureus_: ^^
20:35:16  <piscisaureus_>indutny: it would be nice if you could do this:
20:35:16  <piscisaureus_>fs.watch('foo', callback);
20:35:16  <piscisaureus_>fs.readdir('foo')
20:35:34  <piscisaureus_>indutny: when you do that there should be a guarantee that not events will be lost after readdir
20:37:20  * avalanche123joined
20:38:59  * avalanche123quit (Client Quit)
20:43:11  * paddybyersjoined
20:51:17  * bnoordhuisquit (Ping timeout: 240 seconds)
20:54:19  * kristatequit (Ping timeout: 268 seconds)
21:07:33  * loladiroquit (Quit: loladiro)
21:16:32  * paddybyersquit (Quit: paddybyers)
21:21:28  * mikealquit (Quit: Leaving.)
21:31:21  * mralephquit (Quit: Leaving.)
21:51:01  * avalanche123joined
21:58:01  * hzquit
22:04:05  * avalanche123quit (Quit: Computer has gone to sleep.)
22:08:54  * avalanche123joined
22:10:56  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
22:12:06  * loladirojoined
22:28:20  * avalanche123quit (Quit: Computer has gone to sleep.)
22:35:46  * TooTallNatequit (Quit: Computer has gone to sleep.)
22:45:06  * TooTallNatejoined
22:55:06  * TheJHquit (Quit: goodbye)
23:10:16  * rendarquit
23:11:03  * tjfontainequit (Quit: leaving)
23:14:30  * mmaleckijoined
23:22:48  * avalanche123joined