00:00:07  * brsonquit (Quit: leaving)
00:04:09  * bnoordhuisjoined
00:12:44  * ArmyOfBrucequit (Excess Flood)
00:13:14  * ArmyOfBrucejoined
00:24:05  * AvianFlujoined
00:50:19  * AvianFluquit (Remote host closed the connection)
00:50:48  * AvianFlujoined
00:55:10  * indexzerojoined
00:55:18  * AvianFluquit (Ping timeout: 264 seconds)
01:07:05  * dedis39joined
01:14:02  * loladirojoined
01:15:30  * c4milojoined
01:15:49  * dedis39quit (Remote host closed the connection)
01:21:07  * brsonjoined
01:24:46  * ArmyOfBrucequit (Excess Flood)
01:25:17  * ArmyOfBrucejoined
01:25:34  * indexzeroquit (Quit: indexzero)
01:28:36  * bnoordhuisreimages his freebsd server...
01:28:44  * abraxasjoined
01:33:10  <bnoordhuis>kind of annoying, i'd just finished `make buildworld; make buildkernel` when that security advisory came out
01:47:13  * c4miloquit (Remote host closed the connection)
01:47:39  * c4milojoined
01:51:54  * c4miloquit (Ping timeout: 240 seconds)
03:05:43  <MI6>joyent/libuv: Ben Noordhuis master * 190db15 : unix: set proper loop errno for udp write req cb Harmonize with stream.c - http://git.io/Hq2-qg
03:07:23  * travis-cijoined
03:07:23  <travis-ci>[travis-ci] joyent/libuv#879 (master - 190db15 : Ben Noordhuis): The build passed.
03:07:23  <travis-ci>[travis-ci] Change view : https://github.com/joyent/libuv/compare/665a316aa9d5...190db15638ef
03:07:23  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/libuv/builds/3258676
03:07:23  * travis-cipart
03:44:39  * bradleymeckquit (Quit: bradleymeck)
03:48:06  * chobi_e_joined
03:48:07  * chobi_e_changed nick to chobi_e
03:48:37  * chobi_echanged nick to chobi_e_
03:58:56  * brsonquit (Remote host closed the connection)
03:59:47  * bradleymeckjoined
04:14:56  * ArmyOfBrucequit (Excess Flood)
04:15:25  * ArmyOfBrucejoined
04:15:55  * AvianFlujoined
04:55:53  * ArmyOfBrucequit (Excess Flood)
04:56:20  * ArmyOfBrucejoined
05:12:28  * abraxasquit (Remote host closed the connection)
05:23:06  * abraxasjoined
05:37:24  * AvianFluquit (Remote host closed the connection)
05:47:16  * EhevuTovjoined
05:50:01  * benoitcquit (Excess Flood)
05:50:59  * bradleymeckquit (Quit: bradleymeck)
05:55:55  * philips_quit (Excess Flood)
05:57:12  * philips_joined
05:58:17  * benoitcjoined
06:03:00  * warzquit
06:09:52  * EhevuTovquit (Quit: This computer has gone to sleep)
06:22:04  * btraskjoined
07:47:00  * rendarjoined
07:54:36  * loladiroquit (Quit: loladiro)
08:02:33  * loladirojoined
08:07:04  * loladiroquit (Client Quit)
08:17:17  * EhevuTovjoined
08:25:06  * Raltjoined
08:52:28  * hzjoined
09:32:38  * `3rdEdenjoined
09:42:01  * janjongboomjoined
10:32:36  * abraxasquit (Remote host closed the connection)
10:57:12  * `3rdEdenchanged nick to `3rdEden|BR`b
11:21:09  * EhevuTovquit (Quit: This computer has gone to sleep)
11:38:46  * EhevuTovjoined
11:41:06  * mmaleckichanged nick to mmalecki[out]
11:44:10  * EhevuTovquit (Ping timeout: 252 seconds)
12:12:54  * sgallaghjoined
12:15:52  * abraxasjoined
12:33:31  * bradleymeckjoined
12:38:43  * hzquit (Read error: Connection reset by peer)
12:39:52  * hzjoined
12:40:04  * abraxasquit (Remote host closed the connection)
13:07:45  <sgallagh>bnoordhuis: Could you/have you submitted https://github.com/joyent/node/compare/b61ae54e18c9f6ada7e746052ef6b39272b18a15...988a164cd8820974753ba6a9fc1827d2832c59d4 for upstreaming consideration?
13:19:44  <indutny>sgallagh: I like those
13:19:51  <indutny>and it lgtm
13:19:55  <indutny>but is it ready?
13:21:46  <sgallagh>indutny: These are patches that are being carried in Node's bundled copy of openssl that aren't yet upstream
13:21:56  <sgallagh>I'm asking for Ben to try to get them upstreamed
13:24:20  <sgallagh>bnoordhuis: I asked the Fedora maintainer to look at them, and he replied "generally looking at the patches I don't see anything that's really critical, what's the real problem they are solving that requires these patches to be in?"
13:24:56  <indutny>aah
13:25:00  <indutny>ooh
13:30:01  * bradleymeckquit (Quit: bradleymeck)
13:38:26  * jmar777joined
13:45:37  * travis-cijoined
13:45:37  <travis-ci>[travis-ci] wwwouter/node#3 (master - 57a3b64 : wwwouter): The build is still failing.
13:45:37  <travis-ci>[travis-ci] Change view : https://github.com/wwwouter/node/compare/dfce3ed51482...57a3b646e45b
13:45:37  <travis-ci>[travis-ci] Build details : http://travis-ci.org/wwwouter/node/builds/3263952
13:45:37  * travis-cipart
13:56:10  * `3rdEden|BR`bchanged nick to `3rdEden
13:57:35  * AvianFlujoined
14:20:44  * bradleymeckjoined
14:23:16  * benoitcquit (Changing host)
14:23:16  * benoitcjoined
14:48:07  * c4milojoined
14:57:06  * loladirojoined
15:00:45  * fjakobsjoined
15:01:42  <fjakobs>i see connections not being closed if the response is a 304
15:01:54  <fjakobs>I also don't get an end event for that response
15:02:06  <fjakobs>is that a bug or by design?
15:08:52  <bnoordhuis>fjakobs: more details please?
15:09:52  <fjakobs>i use the node-http-proxy. when the response is a 304 I don't get the "close" event and also the connection is not closed
15:10:28  <bnoordhuis>sgallagh: except for the last patch, those are all from google's "fork", the openssl that chromium ships with
15:11:07  <bnoordhuis>sgallagh: that last one ("fix uninitialized memory access") i've sent upstream
15:11:20  <bnoordhuis>i don't know if it actually landed though
15:11:25  <sgallagh>bnoordhuis: I realize that, but I need justification if I'm going to be able to convince upstream to incorporate them. Also, are they actually necessary?
15:11:32  <fjakobs>https://github.com/nodejitsu/node-http-proxy/blob/master/lib/node-http-proxy/http-proxy.js#L263
15:11:38  <sgallagh>i.e. Can node function without them
15:12:00  <bnoordhuis>sgallagh: no and yes (i.e. not strictly necessary and yes, node functions without them)
15:12:03  * c4miloquit (Remote host closed the connection)
15:12:04  <fjakobs>https://github.com/joyent/node/blob/master/lib/http.js
15:12:10  <sgallagh>ok, good
15:12:12  <fjakobs>line 995
15:12:18  <sgallagh>Then I won't fight over them right now :)
15:12:25  <bnoordhuis>fjakobs: you should ping mmalecki about node-http-proxy
15:12:50  <bnoordhuis>fjakobs: but questions about lib/http.js i can answer :)
15:13:18  <fjakobs>so how are 304 handled differently in node?
15:13:26  <bnoordhuis>fjakobs: they aren't
15:13:29  <fjakobs>I see special handling code there
15:13:42  <fjakobs>I see _hasBody = false
15:14:15  <bnoordhuis>fjakobs: yes, but that is if you're the server sending the 304
15:14:27  * c4milojoined
15:14:36  * piscisaureus_joined
15:14:49  <fjakobs>since i proxy the 304 I am also the server
15:15:21  <bnoordhuis>fjakobs: if your upstream server is sending you a 304 with a response body, it's doing it wrong
15:16:04  <fjakobs>no no, I don't get the response body.
15:16:15  <bnoordhuis>okay, good. so what's the issue?
15:16:34  <fjakobs>but somehow I also don't get a close event for the response
15:16:41  <fjakobs>and the connection stays open
15:17:37  <bnoordhuis>fjakobs: that's WAE (working as expected)
15:18:07  <bnoordhuis>the connection doesn't have to be closed if you receive a 304
15:18:15  <bnoordhuis>and that's why you don't get a close event
15:18:20  <fjakobs>right
15:18:31  <piscisaureus_>is there no "end" event on the response body?
15:18:33  <fjakobs>good to know
15:18:36  <piscisaureus_>(even if there is none)
15:18:56  <bnoordhuis>yes, you get an end event
15:19:40  <fjakobs>my problem is that I want to close the server with "server.close();"
15:19:52  <fjakobs>but it hangs because of the open connection
15:20:02  <fjakobs>even though no request is in preogress
15:21:16  <bnoordhuis>fjakobs: right. you'll have to keep track of that yourself
15:21:28  <bnoordhuis>server.close() just stops the server from accepting new connections
15:21:37  <fjakobs>good to know
15:22:25  <fjakobs>so i will count connections and exit when it drops to 0
15:22:45  <fjakobs>count HTTP requests i mean
15:23:55  <bnoordhuis>fjakobs: yes. and maybe force-close connections that stay open too long
15:24:12  <fjakobs>that helps, thanks
15:24:23  <bnoordhuis>np
15:29:48  * fjakobsquit (Quit: fjakobs)
15:36:44  * bradleymeckquit (Quit: bradleymeck)
15:42:50  * bradleymeckjoined
15:46:47  <creationix>is it my responsibility to call uv_read_stop even though I'm already calling uv_shutdown and uv_close?
15:47:06  <creationix>will either of those clean up any resources that uv_read_stop would have?
15:48:40  <bnoordhuis>creationix: yes
15:48:45  <bnoordhuis>(boolean joke there)
15:49:07  <creationix>that's a relief, at lease a || !a is still true
15:49:20  <bnoordhuis>uv_close() does the same cleanup (and more) as uv_read_stop() does
15:49:22  <creationix>*least
15:49:40  <creationix>ok, so there is no read to call uv_read_stop if I'm sure to always call uv_close?
15:49:46  <creationix>s/read/reason/
15:50:04  * bradleymeckquit (Quit: bradleymeck)
15:50:04  <bnoordhuis>no, indeed
15:51:02  <bnoordhuis>creationix: btw, uv_shutdown() only shuts down the write end
15:51:12  <creationix>yeah, that's what I thought
15:51:14  <piscisaureus_>yeah you should call uv_read_stop. I mean, what if data comes out of a closed socket? That'd be bad.
15:51:15  <bnoordhuis>so depending on what you do, you may need to call uv_read_stop()
15:51:16  <creationix>just wanted clarification
15:52:13  <creationix>how could data come out of a closed socket?
15:52:23  <bnoordhuis>i think piscisaureus_ is trolling for attention :)
15:52:51  * bradleymeckjoined
15:53:10  <piscisaureus_>Isn't that a Tautology ?
15:53:34  <piscisaureus_>s/T/t/
15:54:56  * AvianFluquit (Remote host closed the connection)
15:55:49  <bnoordhuis>you youngsters have forgotten so much about the fine art of trolling
15:56:17  <bnoordhuis>it's not about getting attention, it's about getting the other worked up to furious anger
15:56:34  <bnoordhuis>preferably the kind where they lose it on the internets
15:57:02  <tjfontaine>great vengance and furious anger
16:02:25  * loladiroquit (Quit: loladiro)
16:11:22  <piscisaureus_>bnoordhuis: so almost all buffers are "big" in the end right? every buffer comes from a slab one way or another, unless a user creates a SlowBuffer explicitly
16:11:32  <bnoordhuis>piscisaureus_: yes
16:11:33  <piscisaureus_>bnoordhuis: so mmap()ing seems to be just fine
16:11:39  <piscisaureus_>bnoordhuis: (on unices :-))
16:11:45  <bnoordhuis>yeah, i don't see any issues with it except for speed
16:11:54  <bnoordhuis>but it seems to balance out quite well
16:12:14  <piscisaureus_>does it increase heap fragmentation?
16:12:18  <piscisaureus_>that's also a risk in a way
16:12:26  <bnoordhuis>piscisaureus_: no, because it doesn't use the heap
16:12:34  <piscisaureus_>well - VM fragmentation :-)
16:12:36  <bnoordhuis>now, if you ask "does it increase vm fragmentation"
16:12:45  <piscisaureus_>which I just did :-)
16:12:49  <bnoordhuis>:)
16:12:59  <bnoordhuis>i don't rightly know
16:13:14  <piscisaureus_>bnoordhuis: I think it'd be nice for solaris users too
16:13:22  <piscisaureus_>because we've seen issues with sbrk there
16:13:27  <bnoordhuis>yeah
16:14:00  <bnoordhuis>btw, i landed a follow-up commit that caches the last n mmap'ed regions
16:14:10  <bnoordhuis>which gives it a slight edge over regular new/delete
16:14:37  <piscisaureus_>bnoordhuis: what branch? I only have a sha atm
16:14:44  <bnoordhuis>piscisaureus_: https://github.com/bnoordhuis/node/compare/mmap-buffers
16:16:45  <bnoordhuis>the number 16 was established empirically btw
16:22:02  * philips_quit (Excess Flood)
16:24:15  * philips_joined
16:25:18  * bradleymeckquit (Ping timeout: 264 seconds)
16:26:52  <bnoordhuis>piscisaureus_: bertje, what are you working on this week?
16:29:38  <piscisaureus_>bnoordhuis: dunno yet. atleast, dotjs preso, runjs
16:29:42  <piscisaureus_>fixing windows bugs
16:29:45  <piscisaureus_>bnoordhuis: something else
16:29:53  <bnoordhuis>piscisaureus_: dotjs?
16:30:00  <piscisaureus_>bnoordhuis: conference
16:30:05  <piscisaureus_>in paris
16:30:06  <bnoordhuis>oh right
16:30:18  <piscisaureus_>bnoordhuis: should we "harden" fs routines to always report ENOENT when the filename containes a \0
16:30:32  <piscisaureus_>right now this is sorta a security issue that nobody understands
16:30:47  <bnoordhuis>piscisaureus_: i don't know. it feels like dumbing down things for the unwashed masses
16:31:04  <bnoordhuis>but otoh, it is indeed an issue that few realize exists
16:31:16  <piscisaureus_>I think jan just hacked our service with this
16:31:37  <piscisaureus_>bnoordhuis: also, it isn't really a matter of input sanitization
16:31:46  <bnoordhuis>no doubt. i wouldn't call c9 the epitome of security :)P
16:32:15  <bnoordhuis>i guess i could live with it
16:32:19  <piscisaureus_>bnoordhuis: it's an artifact of how js strings interplay with c string
16:32:20  <piscisaureus_>s
16:32:28  <piscisaureus_>bnoordhuis: also, actually they are very serious about security atm
16:33:28  <bnoordhuis>piscisaureus_: yeah? so the thing i told jan about what i'd do when another security bug was uncovered helped?
16:34:04  <piscisaureus_>bnoordhuis: hahahaha
16:36:37  * loladirojoined
16:36:48  * loladiroquit (Client Quit)
16:37:10  * hzquit
16:37:40  * hzjoined
16:38:40  * `3rdEdenquit (Ping timeout: 246 seconds)
16:38:46  * `3rdEden_joined
16:42:29  <piscisaureus_>bnoordhuis: I have the same feeling about http response splitting btw
16:43:07  <piscisaureus_>bnoordhuis: I dont think anyone who does writeHead(200, { "foo": "bar\r\n\baz: 3" }) intends to write 2 headers
16:45:52  * bradleymeckjoined
16:47:31  <bnoordhuis>piscisaureus_: people should take care. also, i don't want O(n) operations in a hot code path
16:48:17  <piscisaureus_>bnoordhuis: true - we could throw if they put a \n in there :-)
16:48:53  * sblomjoined
16:50:55  <piscisaureus_>Header fields can be extended over multiple lines by preceding each extra line with at least one SP or HT. Applications ought to follow "common form", where one is known or indicated, when generating HTTP constructs, since there might exist some implementations that fail to accept anything
16:51:55  <piscisaureus_>bnoordhuis: so value = value.replace(/\r?\n/g, '\r\n\t') would fix it :-)
16:52:33  <sblom>bnoordhuis: What's the right include for uint64_t? ( https://github.com/joyent/node/pull/4285#discussion_r2158887 )
16:53:10  <bnoordhuis>sblom: stdint.h - but you mentioned msvs 2008 and it might well be broken
16:53:26  <bnoordhuis>okay, off to dinner - back in a few hours
16:53:37  <piscisaureus_>bnoordhuis: sblom: or just include <uv.h>
16:53:45  <piscisaureus_>because uv.h includes it and fixes vs2008
16:55:12  <sblom>piscisaureus_: Cool--and we don't mind a uv "dependency" when there isn't really one? (Just confirming--I don't have a contrary opinion here.)
16:55:25  <piscisaureus_>sblom: well, I don't :-)
16:55:31  <sblom>sold!
16:55:44  <piscisaureus_>sblom: and bnoordhuis is away so we can do whatever we want now
16:56:52  <indutny>bnoordhuis: hoya
16:56:59  <piscisaureus_>indutny: ben is gone
16:57:04  <indutny>bnoordhuis: what the heck with 304s
16:57:10  <indutny>oh
16:57:11  <indutny>right
16:57:42  * dapjoined
16:58:07  <indutny>piscisaureus_: probably you know
16:58:10  <indutny>piscisaureus_: what the heck with 304s
16:58:26  <piscisaureus_>indutny: I think fabian and ben discussed this already.
16:58:33  <piscisaureus_>indutny: but what exactly is your question?
16:58:53  <indutny>piscisaureus_: so why 304s connections should not be closed?
16:59:19  <piscisaureus_>indutny: because there is no need. The connection can still be keepalive.
17:14:58  * kristat__quit (Remote host closed the connection)
17:15:46  <indutny>piscisaureus_: hm...
17:15:50  <indutny>ok
17:25:10  * bradleymeckquit (Quit: bradleymeck)
17:25:11  * joshthecoderjoined
17:26:15  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
17:26:49  * c4miloquit (Remote host closed the connection)
17:30:58  * c4milojoined
17:33:29  * TooTallNatejoined
17:34:43  * TheJHjoined
17:42:22  * AvianFlujoined
17:44:42  * AvianFluquit (Remote host closed the connection)
17:47:27  * ircretaryquit (Ping timeout: 276 seconds)
17:57:57  * piscisaureus_joined
17:59:48  * janjongboomquit (Quit: janjongboom)
18:04:40  * loladirojoined
18:04:41  * loladiroquit (Client Quit)
18:11:47  <piscisaureus_>Why is master so much slower than 0.8
18:11:49  <piscisaureus_>this makes me sad
18:16:37  * bradleymeckjoined
18:20:09  * loladirojoined
18:21:12  <joshthecoder>piscisaureus_: libuv is on vacation (see topic) ;)
18:21:39  <piscisaureus_>:-p
18:28:33  * janjongboomjoined
18:44:59  * loladiroquit (Quit: loladiro)
18:47:27  * `3rdEden_changed nick to `3rdEden
18:48:09  * c4miloquit (Remote host closed the connection)
18:48:39  * bradleymeckquit (Ping timeout: 244 seconds)
18:48:43  * janjongboomquit (Quit: janjongboom)
18:51:24  * sblomwonders aloud: what does uv stand for?
18:55:36  * EhevuTovjoined
18:56:25  * TheJHquit (Ping timeout: 246 seconds)
18:56:42  * TheJHjoined
18:57:44  * loladirojoined
18:59:47  * EhevuTovquit (Client Quit)
19:03:24  <creationix>Someone remind me what the proper response is when the client sends an http request with an "Expect: 100-continue" header
19:03:43  <creationix>respond with "100 continue" as a response, or as the first line before the real response?
19:04:02  * stagasjoined
19:10:41  * brsonjoined
19:24:50  * loladiroquit (Quit: loladiro)
19:34:17  * hzquit (Read error: Connection reset by peer)
19:35:21  * loladirojoined
19:37:57  * c4milojoined
19:40:26  * bradleymeckjoined
19:41:27  <MI6>joyent/node: piscisaureus created branch nosplit - http://git.io/Yf-lPg
19:41:36  <piscisaureus_>^-- bnoordhuis, indutny, isaacs -> thoughts?
19:41:56  <piscisaureus_>I don't see any measurable performance impact
19:42:10  <piscisaureus_>(although note that master is ~ 5% slower than 0.8 atm)
19:42:52  <MI6>joyent/node: Bert Belder nosplit * 68f0a36 : http: protect against response splitting attacks - http://git.io/jHyrmQ
19:43:10  * loladiro_joined
19:43:29  * loladiroquit (Ping timeout: 246 seconds)
19:43:29  * loladiro_changed nick to loladiro
19:44:19  * loladiroquit (Client Quit)
19:49:48  * jmar777quit (Remote host closed the connection)
19:50:21  * jmar777joined
19:54:31  * jmar777quit (Ping timeout: 246 seconds)
19:57:09  * hzjoined
19:59:19  <indutny>philips_: looking
20:02:05  * seebees1joined
20:02:37  * seebees1quit (Client Quit)
20:04:20  * warzjoined
20:05:55  * EhevuTovjoined
20:13:36  * seebees1joined
20:13:40  * `3rdEdenquit (Remote host closed the connection)
20:15:35  * hzquit (Read error: Connection reset by peer)
20:16:22  * hzjoined
20:20:35  * c4miloquit (Remote host closed the connection)
20:23:28  * loladirojoined
20:23:42  * hzquit
20:25:28  * seebees1quit (Ping timeout: 245 seconds)
20:27:43  * loladiroquit (Client Quit)
20:30:58  * jmar777joined
20:33:05  * mmalecki[out]changed nick to mmalecki
20:38:08  * loladirojoined
20:38:12  * EhevuTovquit (Ping timeout: 248 seconds)
20:39:32  * loladiroquit (Client Quit)
20:48:40  <sblom>piscisaureus_, bnoordhuis: I've pushed changes for all of your review feedback to https://github.com/joyent/node/pull/4285
20:49:12  <sblom>piscisaureus_: I'm still trying to figure out why benchmarking is sucking so bad for my dev box.
20:49:20  <sblom>I'll update with real perf numbers the second I have them.
20:52:48  * loladirojoined
20:53:08  * joshthecoderquit (Quit: Leaving...)
21:02:05  * c4milojoined
21:13:41  <creationix>I'm getting a steady 40,000 reqs/second at a stable 3Mb rss with moonslice + luv :)
21:13:56  <creationix>libuv + luajit is such a fun combo for little network servers
21:14:26  * creationixgoes to compare to latest node version on the same machine
21:15:38  * loladiroquit (Quit: loladiro)
21:16:55  <creationix>node 0.8.14 hovers around 34Mb rss and gets a solid 25,000 reqs/second
21:21:12  <piscisaureus_>sblom: may be security essentials or maybe a firewall
21:21:27  <piscisaureus_>sblom: or maybe some stupid LSP
21:21:43  <piscisaureus_>sblom: it may help to look at the process manager while benchmarking to see which process is consuming the cpu
21:24:44  <sblom>piscisaureus_: sadly, it's ab.exe that's maxing out a cpu
21:24:52  <piscisaureus_>hmm
21:25:01  <piscisaureus_>sblom: maybe copy ab.exe from your laptop :-)
21:25:12  <sblom>interesting idea.
21:25:30  <piscisaureus_>sblom: also, if you look at "netsh winsock show catalog" do you see any layered service providers?
21:25:35  <sblom>I think I got both from the official Apache installer for Windows.
21:25:38  <sblom>Will look.
21:29:34  <creationix>I wonder if wrk runs on windows
21:29:44  <creationix>I find it a lot better than ab on linux and osx
21:30:02  <creationix>sblom, https://github.com/wg/wrk
21:30:08  <sblom>creationix: I'll take a look at that--thanks.
21:30:28  <creationix>looks like you'll have to write your own windows makefile
21:31:24  * sgallaghchanged nick to sgallagh_afk
21:31:55  <piscisaureus_>It looks like we'd have to port it to libuv first
21:32:12  <creationix>is redis's event loop not cross-platform?
21:32:19  <piscisaureus_>no
21:32:32  <piscisaureus_>microsoft did an iocp backend for it tho
21:32:46  <piscisaureus_>that is, igor and henry
21:33:09  <creationix>can the windows backend be used standalone I wonder (like in wrk)
21:33:24  <piscisaureus_>bnoordhuis: sure, still an o(n) scan.
21:33:28  <piscisaureus_>bnoordhuis: but it is not noticable
21:33:53  <piscisaureus_>bnoordhuis: remember that copying out the headers involves a couple of O(n) scans already
21:34:23  <creationix>that's a smaller n though ;)
21:35:06  <piscisaureus_>not a smaller n
21:36:12  <bnoordhuis>piscisaureus_: i'm still against it
21:36:20  <bnoordhuis>let express and friends handle it and take the hit
21:36:35  <piscisaureus_>bnoordhuis: I strongly disagree
21:36:45  <bnoordhuis>i strongly disagree with you strongly disagreeing
21:37:01  <creationix>let's just throw away http.js and let it be user-space
21:37:02  <creationix>:)
21:37:14  <creationix>only provide the http_parser bindings in core
21:37:53  <bnoordhuis>not a half bad idea
21:37:59  <creationix>(I'm actually serious on that, but I doubt it could ever happen given the API stability node wants to have)
21:38:55  <creationix>my http.js replacement is called "web". It's a lot simpler, faster, and easier to compose modules without needing something like connect
21:39:21  <bnoordhuis>the thing i dislike about http.js is the amount of cruft and bolt-on hacks it's assembled over the years
21:39:27  <creationix>exactly
21:41:42  * lohkeyjoined
21:42:32  * bradleymeckquit (Quit: bradleymeck)
21:43:35  * creationixpart ("Leaving")
21:43:40  * creationixjoined
21:54:40  <Raynos>creatonix: https://github.com/creationix/web
22:07:10  * sblomquit (Ping timeout: 246 seconds)
22:08:11  <Raynos>creatonix: oh it's node-web.
22:08:26  * AvianFlujoined
22:11:19  <creationix>Raynos, right, it's the same API style as my lua moonslice project
22:12:20  * rendarquit
22:22:32  * AvianFluquit (Remote host closed the connection)
22:29:38  <MI6>joyent/node: Bert Belder wip * 46141d2 : windows: fix normalization of UNC paths - http://git.io/dVrqdQ
22:33:59  * sblomjoined
22:36:16  * joshthecoderjoined
22:45:43  * loladirojoined
22:48:42  * joshthecoderquit (Ping timeout: 264 seconds)
22:50:16  * AvianFlujoined
22:51:31  * TheJHquit (Ping timeout: 265 seconds)
22:57:52  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
23:04:30  * warzquit
23:06:41  * loladiroquit (Quit: loladiro)
23:11:33  * TooTallNatequit (Quit: Computer has gone to sleep.)
23:36:42  * c4miloquit (Remote host closed the connection)
23:42:38  * piscisaureus_joined
23:46:34  <MI6>joyent/libuv: Ben Noordhuis master * d8a8379 : test: disable fs event test on dragonflybsd Disable the fs_event_close_i (+4 more commits) - http://git.io/mRVN7Q
23:46:46  <bnoordhuis>^ my 700th commit to libuv
23:47:39  <tjfontaine>thank god it was for yet another bsd :)
23:48:17  * travis-cijoined
23:48:17  <travis-ci>[travis-ci] joyent/libuv#881 (master - d8a8379 : Ben Noordhuis): The build passed.
23:48:17  <travis-ci>[travis-ci] Change view : https://github.com/joyent/libuv/compare/190db15638ef...d8a8379751ff
23:48:17  <travis-ci>[travis-ci] Build details : http://travis-ci.org/joyent/libuv/builds/3273104
23:48:17  * travis-cipart
23:49:29  <chilts>bnoordhuis++
23:49:42  <chilts>no bot to keep a tally :(
23:50:24  <tjfontaine>it's just missing right now