00:00:00  * ircretaryquit (Remote host closed the connection)
00:00:07  * ircretaryjoined
00:00:26  * tnorrisjoined
00:00:54  <tnorris>tjfontaine: well, that was fun. just filled mem and swap with buffers. =P
00:01:01  <tjfontaine>SURPRISE OOM!
00:01:01  <LOUDBOT>I HAVE PREGNANCY, PREMATURE EJACULATION, PREMATURE OVARIAN FAILURE, AND PROSTATE GLAND ENLARGEMENT
00:01:51  <tjfontaine>LOUDBOT: you're the best
00:01:51  <LOUDBOT>tjfontaine: LET'S TALK ABOUT THE STACK MORKET
00:02:13  * TooTallNatequit (Ping timeout: 248 seconds)
00:02:31  * trevnorrisquit (Ping timeout: 245 seconds)
00:07:47  * TooTallNatejoined
00:08:16  * AvianFluquit (Remote host closed the connection)
00:10:01  * kevinswiberquit (Remote host closed the connection)
00:11:00  * tnorrisquit (Quit: Leaving)
00:12:43  * bnoordhuisjoined
00:17:20  * bnoordhuisquit (Ping timeout: 252 seconds)
00:19:05  <kellabyte>TooTallNate: do you know how to set where the VS "Debug" and "Release" directories go? I tried -Gintermediate_dir= and -Gshared_intermediate_dir= but didn't make a difference
00:19:18  <kellabyte>TooTallNate: -Goutput_dir works for the final binary though
00:19:51  <TooTallNate>kellabyte: what do you mean "it works for the final binary"?
00:19:52  * kazuponjoined
00:20:54  <kellabyte>TooTallNate: like the .exe, but all the pdb, and other stuff it uses while building is going into ./Debug, ./Release, I'd like to make it like ./build/Debug
00:21:17  <TooTallNate>kellabyte: right, that's how node-gyp does it
00:21:50  <TooTallNate>kellabyte: so you need a combination of --generator-output=/path/to/build and -Goutput_dir=.
00:22:10  <rvagg>TooTallNate: I think we need you in on this discussion: https://github.com/isaacs/npm/issues/1891
00:22:56  <TooTallNate>rvagg: i mean ya… i think at this point everyone thinks binary native addons would be a good thing
00:23:30  <rvagg>yeah, but we need to make it actually happen
00:23:38  <TooTallNate>rvagg: it's getting the physical build farm(s) that's the biggest problem
00:23:49  <TooTallNate>like… someone needs to donate/sponsor that shit
00:24:00  <TooTallNate>like, oh idk, joyent, microsoft…
00:24:04  <rvagg>maybe that's a step we can skip for now, make people supply their own to start with
00:24:29  * kazuponquit (Ping timeout: 252 seconds)
00:24:58  <TooTallNate>well where do these binaries get uploaded to?
00:25:08  <TooTallNate>that's probably the next hurdle
00:25:08  <rvagg>npm would be a logical place
00:25:13  <TooTallNate>indeed
00:25:21  <rvagg>with coded names
00:25:30  <rvagg>a hook in npm itself to find and download if available
00:25:38  <rvagg>otherwise invoke node-gyp
00:26:33  <kellabyte>TooTallNate: yeah, that setup isn't exactly what I was hoping, I don't really want to relocate the .sln or VS projects, just want to relocate the Debug and Release directories
00:27:01  <TooTallNate>kellabyte: why don't you want the .sln etc. files in ./build?
00:28:19  <kellabyte>TooTallNate: thats not a typical VS project structure so it might throw some people off
00:29:20  <TooTallNate>kellabyte: but gyp is generating those files right? so they're essentially… temporary
00:29:31  <kellabyte>TooTallNate: not for people developing on windos
00:29:33  <kellabyte>windows
00:29:51  <TooTallNate>kellabyte: in any case, i'm not sure of how to do what you're trying to do (or if it's possible)
00:29:51  <TooTallNate>rvagg: i mean that sounds good
00:30:07  <TooTallNate>kellabyte: you've lost me
00:30:13  <kellabyte>TooTallNate: okie cool, thanks for making me aware of that option :)
00:30:47  <TooTallNate>rvagg: i'd imagine a confirmation prompt the first time you download a binary from a particular "uploader"
00:31:11  <kellabyte>TooTallNate: its just not an expected structure people who typically work in VS will expect
00:31:18  <TooTallNate>rvagg: "x-native-module binary will be downloaded, uploaded by @rvagg, do you accept?"
00:31:21  <TooTallNate>something like that…
00:31:28  <TooTallNate>acceptions can be cached
00:31:34  <TooTallNate>kinda like SSH
00:31:36  <kellabyte>TooTallNate: but no worries your busy and you gave me an option so forget about it :) thanks so much!
00:32:18  <TooTallNate>kellabyte: no prob! sorry i couldn't help you out more specifically
00:33:03  <kellabyte>TooTallNate: you taught me some more things so not pointless :)
00:33:39  <TooTallNate>kellabyte: i know you're one who respects her knowledge :)
00:34:00  * tjfontainewalks away quietly
00:34:13  <TooTallNate>"addicted to brain stimulation" is a term i've used before :p
00:34:19  * dapquit (Quit: Leaving.)
00:36:42  <isaacs>tjfontaine: the last time we changed it (from a tarball-per-release to a folder-per-release) it was super annoying for those of us who maintain node version managers.
00:39:37  * octetcloudquit (Ping timeout: 245 seconds)
00:41:08  * TooTallNatequit (Quit: Computer has gone to sleep.)
00:56:42  * TooTallNatejoined
01:00:42  * wolfeidaujoined
01:02:34  * loladiroquit (Quit: loladiro)
01:03:04  * inolenquit (Quit: Leaving.)
01:06:52  * abraxasjoined
01:14:08  * piscisaureus_joined
01:24:53  * TooTallNatequit (Quit: Computer has gone to sleep.)
01:26:26  * mikealquit (Quit: Leaving.)
01:33:58  * AvianFlujoined
01:38:14  * piscisaureus_quit (Ping timeout: 272 seconds)
01:43:33  * loladirojoined
01:52:15  * inolenjoined
02:00:44  * indexzerojoined
02:02:25  * pooyaquit (Quit: pooya)
02:07:28  <rvagg>isaacs: you're probably not here but in case you are, do you know if there is a live feed of the joyent event happening now?
02:08:22  * nodeskippyjoined
02:08:59  <rvagg>nmind.. you're up talking now and I'm being told it's being recorded
02:12:33  * nodeskippyquit (Ping timeout: 240 seconds)
02:15:24  * inolenquit (Quit: Leaving.)
02:15:34  * piscisaureus_joined
02:22:16  * piscisaureus_quit (Ping timeout: 256 seconds)
02:24:29  * brsonquit (Ping timeout: 246 seconds)
02:58:38  * c4miloquit (Remote host closed the connection)
02:59:05  * c4milojoined
03:03:48  * c4miloquit (Ping timeout: 264 seconds)
03:10:00  * brsonjoined
03:10:34  * brsonquit (Client Quit)
03:11:28  * brsonjoined
03:30:59  * inolenjoined
03:32:27  * pooyajoined
03:42:52  * pooyaquit (Quit: pooya)
03:50:27  * pooyajoined
03:52:57  * bradleymeckjoined
03:53:03  * bradleymeckquit (Client Quit)
03:58:32  * pooyaquit (Quit: pooya)
04:12:25  * abraxasquit (Remote host closed the connection)
04:15:20  * kenperkinsquit (Quit: Computer has gone to sleep.)
04:18:37  * indexzeroquit (Quit: indexzero)
04:59:53  * indexzerojoined
05:07:29  * abraxasjoined
05:16:50  * indexzeroquit (Quit: indexzero)
05:34:07  * octetcloudjoined
05:37:12  * AvianFluquit (Remote host closed the connection)
05:41:02  * indexzerojoined
05:47:41  * TooTallNatejoined
05:47:43  * TooTallNatequit (Client Quit)
05:57:14  * octetcloudquit (Ping timeout: 264 seconds)
06:24:56  * rendarjoined
06:29:13  * bajtosjoined
06:30:40  * indexzeroquit (Quit: indexzero)
06:40:14  * stolsmajoined
07:01:03  * mikealjoined
07:04:28  * benoitcquit (Excess Flood)
07:05:52  * wolfeidauquit (Remote host closed the connection)
07:07:02  * benoitcjoined
07:16:32  * stagasjoined
07:42:24  * brsonquit (Ping timeout: 272 seconds)
07:49:14  * timoxleyjoined
07:49:45  * wolfeidaujoined
08:08:00  * Benviequit (Read error: Connection reset by peer)
08:08:15  * Benviejoined
08:21:53  <indutny>hoya
08:22:36  <inolen>lo, how was the NY trip?
08:29:26  * bajtosquit (Ping timeout: 256 seconds)
08:29:50  * timoxleyquit (Quit: Computer has gone to sleep.)
08:42:43  * bajtosjoined
08:46:03  * bajtos_joined
08:46:56  * bajtosquit (Read error: Connection reset by peer)
08:47:54  <indutny>kewl
08:47:58  <indutny>beautfiul
08:48:00  <indutny>thanks to hij1nx ;)
08:48:30  * bajtosjoined
08:48:36  * bajtosquit (Remote host closed the connection)
08:48:50  * bajtosjoined
08:49:32  * bajtos__joined
08:52:11  * bajtos___joined
08:52:31  * bajtos_quit (Ping timeout: 245 seconds)
08:53:20  * bajtosquit (Ping timeout: 246 seconds)
08:55:02  * bajtos__quit (Ping timeout: 245 seconds)
09:06:54  * bajtos___quit (Read error: Connection reset by peer)
09:39:27  * bnoordhuisjoined
10:02:00  * abraxasquit (Remote host closed the connection)
10:22:59  * hzjoined
10:54:18  * piscisaureus_joined
10:55:01  * bnoordhuisquit (Ping timeout: 248 seconds)
10:56:07  * bnoordhuisjoined
10:58:49  * loladiroquit (Quit: loladiro)
10:59:10  * kevinswiberjoined
11:06:42  * sgallaghjoined
11:22:47  * AvianFlujoined
11:25:06  * kevinswiberquit (Remote host closed the connection)
11:54:33  * loladirojoined
11:55:46  <bnoordhuis>sigh. up/downgrading openssl is a massive PITA
12:01:29  <piscisaureus_>hi
12:03:30  <bnoordhuis>ho
12:19:40  * kevinswiberjoined
12:53:07  * jmar777joined
12:57:21  <bnoordhuis>isaacs, indutny: i'm considering downgrading openssl in v0.10 to 1.0.0f (the v0.8 version)
12:58:05  <bnoordhuis>several people are reporting these spurious issues that go away as soon as you downgrade. it's probably because of the tls v1.2 thing
13:42:57  * c4milojoined
14:09:21  * qmxquit (Quit: ZNC - http://znc.in)
14:09:51  * qmxjoined
14:10:33  * qmxquit (Changing host)
14:10:34  * qmxjoined
14:10:34  * qmxquit (Excess Flood)
14:12:51  * dominictarrjoined
14:19:56  * qmxjoined
14:20:34  * qmxquit (Client Quit)
14:28:20  * kenperkinsjoined
14:30:57  * piscisaureus_quit (Ping timeout: 252 seconds)
14:41:43  * mikealquit (Quit: Leaving.)
14:47:28  * dapjoined
14:49:49  * octetcloudjoined
14:51:49  * hzquit (Ping timeout: 264 seconds)
14:52:41  * hzjoined
14:55:31  * octetcloudquit (Ping timeout: 258 seconds)
15:01:33  * stagasquit (Read error: Connection reset by peer)
15:10:07  * bajtosjoined
15:17:29  * mikealjoined
15:33:36  <indutny>bnoordhuis: green light from me
15:35:05  <bnoordhuis>indutny: i'm still waiting for the results of the CI to come back :)
15:35:58  <bnoordhuis>i probably need to start it manually but i forgot how :-/
15:36:13  <indutny>heh
15:36:20  <indutny>I have no idea
15:36:26  <indutny>jetlag killed me
15:36:27  <indutny>brb
15:39:21  <creationix>indutny: where did you fly?
15:40:52  * octetcloudjoined
15:43:39  <indutny>moscow
15:43:57  <indutny>was in ny
15:45:02  <creationix>indutny: fun, I just got back to Texas from France.
15:45:23  <indutny>heh
15:46:09  <indutny>nice
15:46:29  <indutny>bnoordhuis: there's some stuff with osx hack in libuv
15:46:32  <indutny>a bug
15:46:41  <indutny>I'll try to figure it out later
15:46:48  <indutny>and I've a test right for it :)
15:47:57  <bnoordhuis>good :)
15:48:05  <bnoordhuis>okay, dinner. biab
15:48:17  <indutny>see ya
15:51:08  * dapquit (Quit: Leaving.)
15:51:46  * qmx_joined
15:52:34  * bnoordhuisquit (Ping timeout: 256 seconds)
15:53:09  * qmx_quit (Changing host)
15:53:09  * qmx_joined
15:53:47  * qmx_changed nick to qmx
16:04:56  * pooyajoined
16:08:40  * TooTallNatejoined
16:16:40  * dominictarrquit (Quit: dominictarr)
16:18:59  * bnoordhuisjoined
16:22:01  * AvianFluquit (Read error: Connection reset by peer)
16:22:28  * AvianFlujoined
16:25:51  <isaacs>bnoordhuis: oh, ok, lgtm
16:25:56  <isaacs>bnoordhuis: ignore my question on the issue, then :)
16:26:25  <isaacs>rvagg: there was not a live stream, because we'll be doing almost the same exact event again in Palo Alto next week
16:26:34  <isaacs>rvagg: so we'll release which ever video looks nicer :)
16:26:54  * dapjoined
16:27:17  * bnoordhuisquit (Ping timeout: 246 seconds)
16:31:16  <tjfontaine>isaacs: are you done updating your scrum?
16:32:09  <isaacs>no, sorry, getting distracted :)
16:33:39  <tjfontaine>hehe
16:33:48  <tjfontaine>didn't expect so
16:34:28  <isaacs>tjfontaine: k, go for it
16:34:35  <tjfontaine>thanks
16:43:19  * kevinswiberquit (Remote host closed the connection)
16:48:41  * TooTallNatequit (Read error: Connection reset by peer)
16:49:01  * kevinswiberjoined
16:50:22  * mikealquit (Quit: Leaving.)
16:50:24  * pooyaquit (Quit: pooya)
16:58:15  * pooyajoined
16:59:21  * dominictarrjoined
17:00:18  * kevinswiberquit (Remote host closed the connection)
17:01:45  * indexzerojoined
17:02:37  * TooTallNatejoined
17:07:40  * indexzeroquit (Quit: indexzero)
17:09:18  <isaacs>oh, right, aesni
17:10:29  <tjfontaine>right
17:11:02  <tjfontaine>we can do it, we just need to float the patch that adds it
17:11:21  <tjfontaine>on the plus side we can use the same for 0.8 and make others happy
17:13:31  * kevinswiberjoined
17:13:38  * nodeskippyjoined
17:13:44  <isaacs>good cal
17:16:41  * brsonjoined
17:19:20  * Kjerskichanged nick to kjerski
17:26:57  * inolenquit (Quit: Leaving.)
17:32:08  * mikealjoined
17:44:11  * KiNgMaRquit (Ping timeout: 245 seconds)
17:50:28  * dominictarrquit (Quit: dominictarr)
17:57:54  * inolenjoined
18:02:40  <TooTallNate>rvagg: isaacs: curious what your thoughts are on https://github.com/isaacs/npm/issues/1891#issuecomment-17089886
18:02:57  <TooTallNate>well, starts here really https://github.com/isaacs/npm/issues/1891#issuecomment-17086026
18:05:17  * russell_hquit (Ping timeout: 246 seconds)
18:06:11  * pfox__quit (Ping timeout: 255 seconds)
18:06:29  * CAPSLOCKBOTquit (Ping timeout: 248 seconds)
18:06:37  * pfox__joined
18:07:01  * LOUDBOTquit (Ping timeout: 248 seconds)
18:08:32  <indutny>isaacs: writev
18:08:36  <indutny>lets fucking land it :)
18:09:07  * inolenquit (Quit: Leaving.)
18:09:26  * inolenjoined
18:13:51  * trevnorrisjoined
18:14:17  * CAPSLOCKBOTjoined
18:14:20  <trevnorris>good afternoon
18:14:27  * LOUDBOTjoined
18:15:14  * nodeskippyquit (Quit: Leaving.)
18:16:03  * amartensjoined
18:18:27  * piscisaureus_joined
18:24:31  * bnoordhuisjoined
18:26:59  * TooTallNatequit (Quit: ["Textual IRC Client: www.textualapp.com"])
18:28:23  <isaacs>indutny: i'll review and land this afternoon
18:28:29  <isaacs>indutny: but i agree. we have consensus on API now.
18:28:32  <isaacs>that was the ahrd part :)
18:29:58  <indutny>yep
18:34:26  <bajtos>bnoordhuis: ping
18:35:51  * kevinswiberquit (Remote host closed the connection)
18:37:20  * mikealquit (Quit: Leaving.)
18:37:25  * brsonquit (Ping timeout: 248 seconds)
18:40:12  * indexzerojoined
18:41:01  * TooTallNatejoined
18:44:06  * brsonjoined
18:45:59  <trevnorris>so a perf on net-raw-pipe says ~50% time spent on copy_user_generic_unrolled.
18:46:50  <trevnorris>hm. says "memory copy with exception handling. This version is for CPUs like P4 that don't have efficient micro code for rep movsq."
18:47:13  * piscisaureus_quit (Ping timeout: 240 seconds)
18:47:37  <tjfontaine>trevnorris: is this clang?
18:47:47  <trevnorris>built w/ clang, yes.
18:47:57  <tjfontaine>might retry with gcc and see if you see the same
18:48:02  <trevnorris>good diea.
18:48:04  <trevnorris>idea
18:48:05  <tjfontaine>well I guess it's in the kernel
18:48:08  * tjfontaineshrugs
18:48:32  <tjfontaine>anything's possible gcc just in case :)
18:49:30  <trevnorris>yeah. that's a crazy amount of time spent copying memory around.
18:52:44  <trevnorris>tjfontaine: same w/ gcc.
18:53:02  <tjfontaine>k didn't figure you'd be that lucky
18:53:26  * trevnorrispart ("Leaving")
18:53:26  <tjfontaine>trevnorris: is this on bare metal or virtual machine?
18:53:26  * piscisaureus_joined
18:53:29  * trevnorrisjoined
18:54:19  <trevnorris>not sure how yet, but want to see what code is causing that to occur.
18:54:51  <bnoordhuis>piscisaureus_: you should review bajtos's debugger PR
18:55:14  <piscisaureus_>bnoordhuis: oh. I have 15 minutes. Do you think that's enough?
18:55:19  <bnoordhuis>trevnorris: copy_user_generic_unrolled is the kernel's memcpy function
18:55:27  <bnoordhuis>memcpy-ish, really
18:55:33  <bnoordhuis>piscisaureus_: probably
18:56:20  <bnoordhuis>trevnorris: it means the kernel is copying in from/out to your process
18:56:48  <bnoordhuis>you probably want to run perf again with -e cycles:u (user cycles only)
18:57:02  <trevnorris>bnoordhuis: i'm just wondering two things. first that code says it'll only be used if ERMS and rep_good aren't supported by the CPU.
18:57:12  <trevnorris>ah, ok. thanks.
18:58:22  <trevnorris>other thing I'm wondering is why that's showing 50% usages.
18:58:34  <trevnorris>bnoordhuis: thanks. the perf report there looks much better.
19:00:34  <bnoordhuis>trevnorris: because you're reading/writing a lot of data
19:00:51  <bnoordhuis>copy_user_*() are called by syscalls like read(), write(), etc.
19:01:46  <trevnorris>ah, ok.
19:02:53  <bnoordhuis>okay, family time again
19:07:30  * bnoordhuisquit (Ping timeout: 256 seconds)
19:09:49  <MI6>joyent/node: Miroslav Bajtoš master * fd9e01c : debugger: print port number when connecting to debuggee (+1 more commits) - https://github.com/joyent/node/compare/45ed54600931...fd9e01c031ac
19:10:18  <MI6>joyent/node: Miroslav Bajtoš v0.10 * 8c2ad47 : debugger: `restart` with custom debug port - http://git.io/UKyAzA
19:16:17  * piscisaureus_quit (Ping timeout: 245 seconds)
19:19:41  * dominictarrjoined
19:23:47  * stagasjoined
19:27:51  * TooTallNatequit (Quit: Computer has gone to sleep.)
19:29:46  * c4miloquit (Remote host closed the connection)
19:30:12  * c4milojoined
19:35:00  * c4miloquit (Ping timeout: 264 seconds)
19:36:40  * trevnorrisquit (Quit: Leaving)
19:38:51  * sgallaghquit (Remote host closed the connection)
19:45:24  * c4milojoined
19:46:04  * KiNgMaRjoined
19:51:41  * bajtosquit (Quit: good night)
19:53:07  <indutny>oh, that was my mistake, actually
19:53:08  <indutny>not a bug
19:53:22  <MI6>nodejs-master: #177 UNSTABLE smartos-x64 (2/583) osx-ia32 (1/583) windows-ia32 (145/583) smartos-ia32 (2/583) windows-x64 (145/583) http://jenkins.nodejs.org/job/nodejs-master/177/
20:13:15  * bnoordhuisjoined
20:13:57  <tjfontaine>bnoordhuis: so I'm not sure what you want to do about the cluster bind twice issue, it cleans up if I leave the a/b.kill() in the grandparent on exit
20:15:38  <saghul>hi guys, should libuv be able to handle a tun device on OSX? would opening the fd with a pipe handle work?
20:15:58  <tjfontaine>this has come up in the past, I don't remember the answer
20:16:58  <saghul>I recall something about the fact that only select could be used, not sure if what indutny did with the select hack would fix it
20:17:13  <indutny>it should work
20:17:27  <saghul>indutny with uv_pipe_open, right?
20:17:37  <indutny>yes
20:17:42  <indutny>let me know if it isn't
20:17:47  <indutny>but you might need fresh v0.10 libuv
20:17:52  <indutny>I've pushed some fixes recently
20:18:04  <saghul>indutny i have a fresh one, as of yesterday :-)
20:18:12  * bnoordhuisquit (Ping timeout: 264 seconds)
20:18:13  <saghul>thanks!
20:19:28  <indutny>great :)
20:19:33  <indutny>I'm in process of writing test for it
20:20:05  * TooTallNatejoined
20:20:53  <tjfontaine>PREFIX=out/dist-osxout/dist-osx/32out/dist-osx/usr/local *sigh*
20:21:18  * pooyaquit (Quit: pooya)
20:23:24  <MI6>nodejs-v0.10: #163 UNSTABLE windows-ia32 (7/581) windows-x64 (8/581) smartos-x64 (2/581) linux-ia32 (1/581) http://jenkins.nodejs.org/job/nodejs-v0.10/163/
20:24:36  * kevinswiberjoined
20:32:39  * jmar777quit (Remote host closed the connection)
20:37:12  * brsonquit (Remote host closed the connection)
20:39:16  * brsonjoined
20:39:33  * russell_hjoined
20:42:08  <indutny>ircretary: tell bnoordhuis to take a look at https://github.com/joyent/libuv/pull/791
20:42:08  <ircretary>indutny: I'll be sure to tell bnoordhuis
20:42:12  <indutny>finally
20:42:16  <indutny>a test for osx select hack
20:57:46  <Raynos>What's the difference between using StringDecoder ( http://nodejs.org/api/string_decoder.html ) and String(buffer) ?
20:59:45  <indutny>utf8
21:03:01  * loladiroquit (Quit: loladiro)
21:04:03  * kazuponjoined
21:05:02  <TooTallNate>indutny: awesome :)
21:05:36  <TooTallNate>indutny: though note that `node < /dev/tty` doesn't work with current node
21:05:46  <TooTallNate>indutny: the REPL hangs without any input being read
21:06:20  <tjfontaine>did you update libuv to test against?
21:06:57  * kazuponquit (Remote host closed the connection)
21:07:34  * kazuponjoined
21:09:28  <TooTallNate>tjfontaine: me?
21:09:42  <tjfontaine>yes you
21:09:47  <tjfontaine>:)
21:10:04  <TooTallNate>tjfontaine: no, but that's just a test case - i don't think the osx select() code has changed in libuv for a while
21:10:09  <TooTallNate>correct me if i'm wrong indutny
21:10:24  <tjfontaine>oh his change previously was only to node?
21:11:08  <tjfontaine>TooTallNate: I thought this is what he did to fix the `< /dev/tty` https://github.com/joyent/libuv/pull/789
21:11:25  <tjfontaine>oh I guess that was macvim
21:11:40  <TooTallNate>tjfontaine: well i appears i'm wrong, possibly
21:11:54  <indutny>TooTallNate: libuv update needed
21:11:57  <TooTallNate>tjfontaine: i guess indutny heard when i was complaining about < /dev/tty the other day :p
21:12:00  <TooTallNate>indutny: <3
21:12:02  <tjfontaine>:)
21:12:05  <TooTallNate>carry on
21:12:09  <indutny>update libu
21:12:13  <indutny>s/libu/libuv
21:12:16  <indutny>update it
21:12:17  <indutny>just
21:12:18  <indutny>update
21:12:20  <indutny>:)
21:12:27  <TooTallNate>Just Do It™
21:12:31  <indutny>haha
21:12:31  <indutny>yeah
21:12:40  <indutny>so basically, you should try updating it
21:12:44  <tjfontaine>ben and I discussed that we should change the libuv update policy for node
21:12:45  <indutny>and pulling that last patch
21:12:46  <indutny>https://github.com/joyent/libuv/pull/791
21:13:00  * dominictarrquit (Quit: dominictarr)
23:16:24  <isaacs>slurp: welcome back
23:16:41  <isaacs>well, that's frustrating.. was eager to paste the link to the logs into that issue. oh well.
23:17:30  <isaacs>tjfontaine, TooTallNate, indutny: By the way, if you ever notice that logs are not logging, you can ssh into root@logs.libuv.org and `svcadm restart /site/slurp` to give it a slap.
23:17:58  <TooTallNate>isaacs: how do we know if logs are being logged?
23:18:05  <tjfontaine>isaacs: ok
23:18:09  <TooTallNate>oh, by looking at them :)
23:18:19  <isaacs>TooTallNate: sure, that, or if "slurp" is not around :)
23:18:26  <isaacs>ircretary: maybe you should start logging things again.
23:18:26  <ircretary>isaacs: I'm not sure what to do with that command. Ask for help in PM.
23:18:27  <TooTallNate>isaacs: ahhh, that too
23:18:31  <isaacs>ircretary: you're fired.
23:18:55  <isaacs>anyway..
23:19:23  <isaacs>The third item, moving tarballs out of couchdb, really has to happen before we serve binaries, because they're freakin gigantic
23:19:26  * timoxleyquit (Quit: Computer has gone to sleep.)
23:20:06  <isaacs>as it is now, the registry.couch file is 64GB, about 90% or more of which is gzipped tarball data, and it's growing at a faster rate each day
23:20:47  <isaacs>binary publishing would only bloat about 1-10% of all packages (since most packags are not binaries), but those packages would bloat by about 10*(number of targets)
23:21:21  * defunctzombie_zzchanged nick to defunctzombie
23:21:28  <isaacs>so, by my rough back-of-the-napkin math, we'd be talking about roughly doubling the size of the registry. Item #3 is a priority in its own right, because replicating the registry is fucking torture, but with binaries, it becomes even more important.
23:21:38  <isaacs>(similar to package signing: important now, WAY MORE important with binaries)
23:21:50  <tjfontaine>agreed
23:22:23  <isaacs>sorry, 10*(number of targets)*(number of published versions)
23:22:32  <isaacs>so, maybe a bit more than double. maybe like 10x
23:22:46  <isaacs>manta can do this. couch cannot.
23:22:53  <isaacs>we're already pushing the limits.
23:23:22  <isaacs>couch shines at "zillions of json docs" not at "a million giant files"
23:31:23  * mikealjoined
23:32:19  * bnoordhuisjoined
23:34:03  <kevinswiber>Is there a GitHub issue outlining these goals? Package signing would be useful for anyone hosting Node apps. As in... they could have a policy of only trusting certain sources. I'd be interested in exploring that.
23:35:00  <tjfontaine>kevinswiber: https://github.com/isaacs/npm/issues/1891 is the one TooTallNate and I were discussing, but it doesn't yet have isaacs thoughts on signing
23:35:08  <TooTallNate>isaacs: also, i don't think we need to upload the entire module for binary tarballs, we just need the "build" dir
23:35:26  <TooTallNate>isaacs: and on top of that, we can filter out known stuff that node doesn't need, like the intermediate .o files
23:35:31  <kevinswiber>tjfontaine: Thanks!
23:35:50  <isaacs>TooTallNate: i actually think that binary modules shouldn't be uploaded by authors
23:35:58  <isaacs>TooTallNate: that just encourages "works on my machine"-ism
23:36:06  <isaacs>TooTallNate: also, then we'll only get linux and os x binaries :)
23:36:25  <TooTallNate>isaacs: interesting… so you want a build farm instead?
23:36:46  <TooTallNate>isaacs: i mean i figured something like that would probably emerge on top of whatever we decide on
23:38:20  <tjfontaine>isaacs: there are enough disparate platforms out there that I don't see that as being maintainable
23:38:41  <tjfontaine>it's not like there's the OneTrueLinux out there
23:39:14  <isaacs>tjfontaine: actually, fuck linux
23:39:18  <tjfontaine>haha
23:39:23  <isaacs>tjfontaine: i mean, not in the sense that we stop supporting it
23:39:32  <tjfontaine>do we get to say the same about windows? :)
23:40:04  <isaacs>but like, if a distro wants to get on board with this, then canonical or redhat or whoever can chat with us and throw some resources at it
23:40:25  <isaacs>windows is actualy one of the easiest to support, and probably about the highest payoff.
23:40:33  <isaacs>but we should do smartos first.
23:40:43  <TooTallNate>O.o
23:40:45  * kazuponjoined
23:40:51  <isaacs>i mean, the thing about linux is that it's one of the hardest, and linux people are pretty good at installing stuff anyway.
23:40:57  <tjfontaine>smartos is certainly the most reliably achievable of the 4 major we support
23:41:04  <isaacs>eah
23:41:28  * kazuponquit (Read error: Connection reset by peer)
23:41:37  <isaacs>smartos is the easiest to do (with joyent's resources anyway) and i figure we cah flesh out the required bits there, and then extend for windows, and then maybe some linuxes.
23:41:54  <isaacs>if we can get microsoft to provide the actual windows build stuff, that'd be good.
23:41:58  <isaacs>but we'll have to tell htem exactly what to do
23:42:04  <isaacs>have clear interfaces for it, etc.
23:42:43  <trevnorris>bnoordhuis: in your opinion, do we care about protecting users from doing things like "Buffer.poolSize = Infinity"?
23:46:05  <TooTallNate>isaacs: that all seems like a lot more work and less flexibility than a centralized npm command :\
23:46:37  <isaacs>trevnorris: i think it's safe to blow up there, but it's also safe to just say "Don't do that"
23:46:58  <isaacs>TooTallNate: well, it's more work for us :)
23:47:19  <isaacs>TooTallNate: but a "you build this and upload it" has pretty deep fundamental problems
23:47:45  <isaacs>TooTallNate: unless gyp can compile a windows and linux and sunos binary on a mac laptop
23:47:53  * kevinswiberquit (Remote host closed the connection)
23:48:28  <isaacs>TooTallNate: also, we still can't even really talk about this until we get package signing and move the tarballs out of couch
23:48:36  <isaacs>TooTallNate: which are both *much* smaller problems anyway :)
23:48:54  <wolfeidau>isaacs: The Linux that needs binary modules the most is ARM, as you say the others are not really stressed
23:49:09  <TooTallNate>well ok then, i'll withdraw…
23:49:22  <bnoordhuis>trevnorris: i don't
23:49:34  <trevnorris>coolio.
23:49:55  <isaacs>TooTallNate, wolfeidau: That's a great point. ARM is pretty awful to compile on.
23:50:24  <bnoordhuis>but also to compile _for_
23:50:29  <TooTallNate>isaacs: it's also the most diverse / least predictable… this is tjfontaine's concern
23:50:54  <TooTallNate>which i think is perfectly valid
23:51:29  * toothrchanged nick to toothrot
23:51:30  <tjfontaine>there is a general need for improving node/npm/node-gyp logic for cross-compiling as well, least of whicih is what TooTallNate hit today
23:51:58  <TooTallNate>tjfontaine: i mean i don't think there's a *need*. It's nice to have though…
23:52:24  <tjfontaine>TooTallNate: joyent needs it in places, not necessarily for arch cross compiling, but for when they're building stuff that's targetting a new platform
23:55:41  <wolfeidau>TooTallNate, tjfontaine it would also speed up large scale deployments using node and npm correct? especially on the m1.micro's of the hosting world :)
23:56:49  * dominictarrquit (Quit: dominictarr)
23:58:13  <tjfontaine>wolfeidau: ya, this will "fix" the kids doing -j4 on micros (lame people)
23:58:15  <tjfontaine>it's c++ folks.
23:58:31  <tjfontaine>that means the ld will use 3x more memory than you have :P
23:58:31  <wolfeidau>LOL
23:59:31  <tjfontaine>https://github.com/joyent/node/issues/5322