00:00:01  * ircretaryquit (Remote host closed the connection)
00:00:09  * ircretaryjoined
00:00:27  * brsonquit (Ping timeout: 260 seconds)
00:04:02  * xaqjoined
00:09:06  * c4milojoined
00:18:57  * kazuponjoined
00:25:32  * kazuponquit (Ping timeout: 255 seconds)
00:25:36  * bradleymeckjoined
00:27:08  * c4miloquit (Remote host closed the connection)
00:28:39  * brsonjoined
00:28:54  * c4milojoined
00:38:34  * bradleymeckquit (Quit: bradleymeck)
00:46:08  * bradleymeckjoined
01:09:33  * mmaleckichanged nick to mmalecki[zzz]
01:15:51  * c4miloquit (Remote host closed the connection)
01:16:21  * xaqquit (Remote host closed the connection)
01:16:26  * c4milojoined
01:17:24  * benoitcquit (Excess Flood)
01:17:54  * benoitcjoined
01:21:11  * c4miloquit (Ping timeout: 260 seconds)
01:21:34  * kazuponjoined
01:25:24  * bnoordhuisquit (Ping timeout: 245 seconds)
01:26:19  * kazuponquit (Ping timeout: 260 seconds)
01:37:00  * abraxasjoined
01:40:09  * xaqjoined
02:02:21  * qmx|awaychanged nick to qmx
02:13:22  * AvianFlu_joined
02:13:30  * AvianFluquit (Disconnected by services)
02:13:37  * AvianFlu_changed nick to AvianFlu
02:25:08  * indutnyquit (Ping timeout: 435 seconds)
02:25:20  * benoitcquit (*.net *.split)
02:25:21  * kazuponjoined
02:25:30  * indutny_joined
02:26:33  * kazuponquit (Ping timeout: 240 seconds)
02:43:06  * russell_hquit (Remote host closed the connection)
02:43:09  * joshthecoderquit (Write error: Broken pipe)
02:43:14  * dsantiag_joined
02:43:16  * benoitcjoined
02:43:20  * dsantiagoquit (Ping timeout: 264 seconds)
02:43:22  * txdv_joined
02:43:23  * joshthecoder_joined
02:46:09  * russell_hjoined
02:49:32  * txdvquit (*.net *.split)
03:01:38  * abraxasquit (Remote host closed the connection)
03:02:40  * abraxasjoined
03:07:12  * indexzerojoined
03:07:38  * loladiroquit (Quit: loladiro)
03:22:47  * kazuponjoined
03:24:26  * loladirojoined
03:25:28  * brsonquit (Quit: leaving)
03:29:03  * kazuponquit (Ping timeout: 252 seconds)
03:46:26  * qmxchanged nick to qmx|away
03:52:01  * xaqquit (Remote host closed the connection)
04:00:44  * bradleymeckquit (Quit: bradleymeck)
04:01:04  * defunctzombie_zzchanged nick to defunctzombie
04:02:47  * bradleymeckjoined
04:09:11  * abraxasquit (Remote host closed the connection)
04:11:46  * bradleymeckquit (Quit: bradleymeck)
04:12:21  * brsonjoined
04:23:04  * bradleymeckjoined
04:25:21  * kazuponjoined
04:30:06  * kazuponquit (Ping timeout: 252 seconds)
04:36:03  * kazuponjoined
04:55:14  * bradleymeckquit (Quit: bradleymeck)
05:00:19  * kazuponquit (Remote host closed the connection)
05:06:16  * abraxasjoined
05:16:41  * bradleymeckjoined
05:35:57  * benoitcquit (Excess Flood)
05:36:07  * trevnorrisjoined
05:39:59  * benoitcjoined
06:00:42  * kazuponjoined
06:05:23  * kazuponquit (Ping timeout: 260 seconds)
06:09:07  * stagasjoined
06:32:47  * mikealquit (Quit: Leaving.)
06:33:08  * mikealjoined
06:40:34  * defunctzombiechanged nick to defunctzombie_zz
06:49:47  * defunctzombie_zzchanged nick to defunctzombie
07:01:24  * kazuponjoined
07:06:02  * kazuponquit (Ping timeout: 255 seconds)
07:15:44  * brsonquit (Ping timeout: 246 seconds)
07:34:39  * defunctzombiechanged nick to defunctzombie_zz
07:35:53  * hzjoined
07:39:08  * kazuponjoined
07:57:55  * mmalecki[zzz]changed nick to mmalecki
08:44:34  * mmaleckichanged nick to mmalecki[fly]
08:54:51  * indexzeroquit (Quit: indexzero)
09:16:23  * rendarjoined
09:21:03  * txdv_changed nick to txdv
09:44:12  * loladiroquit (Quit: loladiro)
09:56:45  * hzquit
10:02:57  * trevnorrisquit (Quit: Leaving)
10:22:49  * kazuponquit (Remote host closed the connection)
10:24:00  * bradleymeckquit (Quit: bradleymeck)
10:31:33  <MI6>joyent/node: Mitar v0.10 * e5b90a1 : querystring: Removing unnecessary binding Binding of `http_parser` in qu - http://git.io/W4yAew
10:31:52  <indutny_>morning
10:32:15  * indutny_changed nick to indutny
10:41:41  * kazuponjoined
10:47:54  <MI6>nodejs-v0.10: #95 UNSTABLE windows-x64 (5/570) smartos-ia32 (1/570) windows-ia32 (4/570) http://jenkins.nodejs.org/job/nodejs-v0.10/95/
10:53:06  * xaqjoined
11:00:48  * abraxasquit (Remote host closed the connection)
11:40:35  * sgallaghjoined
11:49:26  * stagasquit (Ping timeout: 246 seconds)
12:29:01  * piscisaureus_joined
12:53:26  * kazuponquit (Remote host closed the connection)
12:54:43  * AvianFluquit (Remote host closed the connection)
13:02:18  * qmx|awaychanged nick to qmx
13:07:22  * kazuponjoined
13:25:33  * kazuponquit (Remote host closed the connection)
13:27:00  * kazuponjoined
13:43:53  * piscisaureus_quit (Ping timeout: 245 seconds)
14:05:23  * xaqquit (Remote host closed the connection)
14:08:51  * dominictarrjoined
14:09:08  <indutny>hey people
14:13:04  * piscisaureus_joined
14:17:03  * xaqjoined
14:21:39  <indutny>piscisaureus_: hey man
14:21:47  <indutny>wanna take a look at something interesting? ;)
14:21:58  * xaqquit (Ping timeout: 240 seconds)
14:35:35  <indutny>mraleph: hey man
14:35:36  <indutny>yt?
14:36:22  <mraleph>yes
14:36:29  <mraleph>hi
14:36:44  <indutny>mraleph: do you have a minute?
14:36:55  <mraleph>sure
14:37:30  * mikealquit (Quit: Leaving.)
14:37:33  <indutny>mraleph: what do you think about this https://gist.github.com/indutny/afdf1cf3d31d2725dda9
14:38:25  * hzjoined
14:38:26  <mraleph>what is the purpose of this?
14:38:51  <indutny>mraleph: external marking
14:39:11  <indutny>this might help node.js improve buffer performance
14:39:28  * piscisaureus_quit (Read error: Operation timed out)
14:39:33  <indutny>by moving from Persistent+WeakCallback to this marking mechanism
14:39:40  <indutny>and doing our own GC in GCEpilogueCallback
14:39:49  * dominictarrquit (Quit: dominictarr)
14:40:44  <mraleph>I don't understand how is this intended to be used and why it is better than MakeWeak.
14:41:24  * qmxchanged nick to qmx|afk
14:41:27  <indutny>mraleph: well
14:41:32  <indutny>its just external gc
14:41:41  <indutny>and external managed memory
14:42:05  <indutny>we may allocate buffers
14:42:07  * stagasjoined
14:42:08  <indutny>and manage them
14:42:25  <indutny>instead of registering all that crap in isolate
14:42:31  <indutny>which is not really fast
14:42:44  <indutny>this is an experiment
14:42:49  <indutny>not sure if it'll be really faster
14:42:55  <indutny>but I just decided to try it out
14:43:27  <indutny>also it could be used to eliminate slice-retention problem
14:43:35  <mraleph>so why registering "crap" in isolate is not fast?
14:43:38  * loladirojoined
14:44:07  <mraleph>what are you going to do with those that are not marked in your approach?
14:44:14  <mraleph>do they just die always?
14:44:33  <indutny>yeah
14:45:15  <mraleph>like they never ever need to be kept alive because somebody holds a ref to them?
14:45:24  <mraleph>outside of V8 heap.
14:45:33  <indutny>yes
14:46:06  <indutny>that should work
14:46:40  <indutny>I think some sort of mark-region implementation will be good for this purposes
14:46:41  <indutny>but anyway
14:46:51  <indutny>another thing I wanted to ask you
14:47:05  <indutny>is there any other places where I should invoke foreign callback form?
14:47:08  <indutny>s/form/from/
14:47:11  <mraleph>I think you just need a lightweight weak references then. This seems to be too intrusive.
14:47:35  <indutny>mraleph: isn't it the same?
14:47:46  <indutny>why intrusive?
14:48:06  <mraleph>this one punishes the marker, ask you are constantly checking instancetype.
14:48:12  <mraleph>s/ask/as/
14:48:22  <indutny>ah, well
14:48:25  <indutny>yep, this is a bit problem
14:48:28  <mraleph>branch will always be mispredicted, etc.
14:48:35  <indutny>well, it might be fixed
14:48:36  <mraleph>this is hot place.
14:48:43  <indutny>in visitor
14:48:48  <indutny>this is just some preliminary version
14:48:53  <indutny>to demonstrate that its working
14:49:09  <mraleph>yes, it can be moved by moving to lightweight weak references :-)
14:49:19  <indutny>mraleph: do they exist?
14:49:21  <mraleph>s/can be moved/can be fixed/
14:49:30  <mraleph>no, but they can be implemented.
14:49:59  <indutny>ok...
14:50:07  <indutny>but
14:50:15  <indutny>won't be they implemented as a global list
14:50:21  <indutny>as persistent is right now
14:50:27  <indutny>I'm not sure what we'll get from it then
14:51:20  <mraleph>I seriously doubt that having a global list is the thing that causes you problems. Did you actually benchmark/profile that?
14:51:37  <indutny>mraleph: I didn't, but trevnorris did
14:52:04  <indutny>but that's not the only thing that I'm trying to fix
14:52:07  <mraleph>his last mail to v8-users says that he overestimated cost of persistent.
14:52:13  <indutny>ah
14:52:30  <indutny>I'm not subscribed
14:52:33  <indutny>only v8-dev@
14:52:35  <indutny>gosh
14:52:40  <indutny>ok, anyway
14:52:55  <indutny>I want Buffer.slice() results to not retain parent buffer
14:53:03  <indutny>and this possible with external marking
14:53:18  <indutny>and not really with weak references
14:53:44  <indutny>considering that TypedArrays support in v8 is very preliminary
14:53:52  <indutny>and right now even worse than our support of Buffers
14:54:02  <indutny>it might be not that bad thing
14:54:54  <indutny>hm...
14:54:57  <indutny>I need to reconsider it
14:55:09  <mraleph>well, it is possible with weak references as well. in GC epilogue you just iterate slices and check if anybody needs to be "materialized" because parent wrapper died.
14:55:43  <indutny>mraleph: yeah, I believe you
14:55:55  <indutny>this doesn't seem to be worth it without trevor norris stuff
14:56:09  <indutny>because if global list is fast enough
14:56:20  <indutny>external GC won't make things faster
14:56:58  <mraleph>https://groups.google.com/d/msg/v8-users/6kSAbnUb-rQ/wb2ALFMlOv0J
14:57:07  <indutny>yeah, I found it
14:58:10  <creationix>was someone trying to message me? Colloquy says I had a mention, but my scroll back isn't long enough to see it.
14:58:31  <indutny>creationix: I believe bnoordhuis was asking you if you're the owner of gypified account
14:58:41  <indutny>github account
14:58:45  <creationix>never heard of it
14:58:52  <indutny>ok, that wasn't you then :)
14:58:59  <indutny>http://github.com/gypified
15:00:16  <indutny>mraleph: thank you for looking :)
15:00:23  <mraleph>np.
15:00:38  <mraleph>I am always happy to shut some ideas down with my shotgun :-)
15:00:49  <indutny>haha
15:00:53  <indutny>its not completely dead yet
15:01:04  <indutny>I might implement super-light weak references
15:01:07  <indutny>API
15:01:17  * loladiroquit (Quit: loladiro)
15:01:18  <indutny>to avoid mentioned problems
15:01:36  <indutny>ah, btw
15:01:41  <indutny>have I covered all possible cases?
15:01:47  <indutny>i.e. mark-compact visitor and evacuation?
15:01:59  <indutny>or there're any other cases when object should be marke
15:02:02  <indutny>s/marke/marked
15:02:12  <indutny>mraleph: ^
15:02:14  <mraleph>incremental marking has it's own marker.
15:02:18  <indutny>oh gosh
15:02:19  <mraleph>so you should check that out.
15:02:22  <indutny>ok
15:02:23  <indutny>thanks
15:02:27  <indutny>is that all?
15:02:38  <mraleph>I'd say yes.
15:02:42  <indutny>kewl
15:05:19  <mraleph>ah also generated code for incremental write-barrier marks some objects black right away, but I don't think it does that for Foreign.
15:06:37  <indutny>oh gosh
15:06:44  <indutny>well
15:06:49  <indutny>they should not be affected by write-barrier
15:07:02  <indutny>they're not containing any references
15:08:20  <mraleph>incremental write-barrier marks some white objects that have no internal references (e.g. sequential strings, numbers) as black when you store them into a black object to save time on scanning.
15:09:38  <indutny>ah
15:09:48  <indutny>ok
15:14:47  * loladirojoined
15:15:50  * kazuponquit (Remote host closed the connection)
15:16:47  * defunctzombie_zzchanged nick to defunctzombie
15:16:58  * dominictarrjoined
15:19:23  * c4milojoined
15:20:54  * c4miloquit (Remote host closed the connection)
15:21:22  * c4milojoined
15:21:41  * c4miloquit (Read error: No route to host)
15:22:05  * c4milojoined
15:23:30  * mikealjoined
15:24:01  * mmalecki[fly]changed nick to mmalecki
15:24:13  * dominictarrquit (Quit: dominictarr)
15:42:02  * dapjoined
15:46:13  * kazuponjoined
15:48:13  * mikealquit (Quit: Leaving.)
15:52:04  * inolenquit (Quit: Leaving.)
15:54:04  * bnoordhuisjoined
15:54:47  * kazuponquit (Ping timeout: 255 seconds)
16:03:19  * loladiroquit (Quit: loladiro)
16:09:08  * bradleymeckjoined
16:30:12  * indexzerojoined
16:32:53  * brsonjoined
16:33:10  * inolenjoined
16:43:49  * indexzeroquit (Quit: indexzero)
16:44:15  * dominictarrjoined
16:44:24  <MI6>joyent/node: Kyle Robinson Young master * 889fec3 : doc: typo fixes - http://git.io/Peh3PQ
16:48:35  * TooTallNatejoined
16:49:05  * mikealjoined
16:51:29  * kazuponjoined
16:52:27  * qmx|afkchanged nick to qmx
16:53:45  * TooTallNatequit (Quit: ["Textual IRC Client: www.textualapp.com"])
16:57:39  <MI6>nodejs-master: #133 FAILURE osx-x64 (1/573) osx-ia32 (1/573) smartos-ia32 (3/573) smartos-x64 (2/573) http://jenkins.nodejs.org/job/nodejs-master/133/
16:58:20  <tjfontaine>right
16:59:11  * bnoordhuisquit (Ping timeout: 246 seconds)
17:00:28  <MI6>joyent/node: Timothy J Fontaine master * b3a8e0d : etw: update prototypes to match dtrace provider The DTrace probes were u - http://git.io/5F45Mg
17:01:01  <tjfontaine>which one of you nice folk merged that? :)
17:01:19  * sgallaghquit (Remote host closed the connection)
17:02:12  * sgallaghjoined
17:02:27  <isaacs>tjfontaine: piscisaureus committed 3 minutes ago
17:02:31  <isaacs>https://github.com/joyent/node/commits/
17:03:05  <tjfontaine>oh I just looked at that github commit link
17:09:37  <MI6>joyent/node: Michael Hart v0.10 * 440dcae : Ensure BAD domain example actually uses domain - http://git.io/bAG28w
17:10:56  <isaacs>indutny: review, please? https://github.com/joyent/node/pull/5170
17:11:12  <isaacs>indutny: i can't figure out how to easily reproduce, though.
17:11:32  <isaacs>indutny: since any sane TLS implementation will close the connection when the TLS bit is ended.
17:11:49  <isaacs>indutny: but i guess if IIS doesn't, and we don't, then both parties have the bug, and all hell breaks loose?
17:17:39  * indexzerojoined
17:19:26  <MI6>nodejs-master: #134 UNSTABLE linux-ia32 (1/573) windows-ia32 (5/573) windows-x64 (5/573) smartos-ia32 (4/573) smartos-x64 (5/573) http://jenkins.nodejs.org/job/nodejs-master/134/
17:23:11  <bradleymeck>anyone have an example of implementing a new type of uv_stream_t as a third party lib? I see bunch of stuff with PRIVATE around it
17:23:59  * AvianFlujoined
17:24:08  * piscisaureus_joined
17:25:04  * benoitcquit (Excess Flood)
17:25:11  * benoitcjoined
17:32:11  <indutny>isaacs: hoya
17:32:30  <indutny>isaacs: test?
17:32:33  <indutny>otherwise LGTM
17:32:45  <isaacs>indutny: i'm not sure how to test it
17:32:47  <indutny>isaacs: TLS bit isn't ended
17:32:51  <indutny>isaacs: aaah
17:32:54  <indutny>that's IIS thing
17:32:55  <indutny>sorry
17:32:59  <isaacs>indutny: everything i can try to throw at it still passes without the patch
17:33:10  <isaacs>indutny: because our tls *server* is not busted like IIS
17:33:13  <isaacs>indutny: but our client was.
17:33:14  <indutny>isaacs: haha
17:33:26  <indutny>isaacs: LGTM
17:33:33  <indutny>but I think that should be fixed on net.js level too
17:33:38  <indutny>probably in 0.12
17:33:47  * eris0xffjoined
17:33:58  <eris0xff>hi all
17:34:01  <indutny>hi
17:34:11  <isaacs>indutny: fixed in net.js level?
17:34:12  <eris0xff>whats the proper forum for suggesting an API call?
17:34:13  <isaacs>indutny: howso?
17:34:20  <indutny>isaacs: well, remember my patch?
17:34:23  <indutny>isaacs: about halfOpen
17:34:27  <isaacs>eris0xff: https://github.com/joyent/node/issues
17:34:32  <isaacs>indutny: oh, that.
17:34:35  <indutny>isaacs: yes
17:34:36  <tjfontaine>or libuv/issues depending
17:34:37  <eris0xff>(suggesting adding it?)
17:34:40  <indutny>that's a *proper* fix
17:34:44  <indutny>but it breaks other things
17:34:46  <tjfontaine>eris0xff: yes
17:34:51  <eris0xff>ok thx
17:34:53  <isaacs>indutny: well, it breaks other things because it's not a proper fix :)
17:35:07  <indutny>isaacs: well... not really
17:35:19  <indutny>it follows paradigm behind half-open option
17:35:27  <indutny>connection shouldn't be alive if we closed it
17:35:29  <indutny>and shutdown it
17:35:31  <tjfontaine>isaacs: any problem with me closing tickets where I know the feature has landed?
17:36:01  <isaacs>tjfontaine: nope. extra credit to add the commit reference, but sometimes i just do "Landed in v0.10/master/whatever"
17:36:05  <tjfontaine>nod
17:38:24  * trevnorrisjoined
17:38:57  <MI6>nodejs-v0.10: #96 UNSTABLE linux-x64 (2/570) windows-x64 (4/570) smartos-x64 (5/570) smartos-ia32 (3/570) windows-ia32 (4/570) http://jenkins.nodejs.org/job/nodejs-v0.10/96/
17:40:08  <trevnorris>with "Buffer* Buffer::New(const char* data, size_t length)" it's possible for a developer to give "length" longer than "char*", correct?
17:40:51  <indutny>trevnorris: hey man
17:40:58  <trevnorris>sup?
17:41:05  <indutny>trevnorris: wanna experiment a bit?
17:41:12  <indutny>https://gist.github.com/indutny/afdf1cf3d31d2725dda9
17:41:22  <indutny>its a bit rough, but may give us an idea
17:41:26  <indutny>err
17:41:28  <indutny>not idea
17:41:39  <indutny>anyway
17:42:21  <trevnorris>awesome. give me a few to understand it
17:42:30  <bradleymeck>trevnorris: i would assume you can tell it a larger length than allocated
17:43:00  <Ralt>hi people
17:43:06  <trevnorris>bradleymeck: just wanted to confirm. so it would be on the developer to make sure they don't.
17:43:34  <bradleymeck>yup
17:43:43  <trevnorris>cool. thanks
17:43:46  <Ralt>is there some wiki giving some guidelines about which way you'd appreciate people to contribute to joyent/node?
17:44:14  <trevnorris>Ralt: https://github.com/joyent/node/wiki/Contributing
17:44:21  <trevnorris>Ralt: https://github.com/joyent/node/wiki/Contributing-for-Dummies
17:44:33  <Ralt>heh. Perfect.
17:44:39  <Ralt>thanks
17:44:43  <trevnorris>np
17:45:39  <tjfontaine>there's also its companion https://github.com/joyent/node/blob/master/CONTRIBUTING.md
17:50:16  * qmxchanged nick to qmx|lunch
17:51:34  * bnoordhuisjoined
17:53:44  <trevnorris>indutny: awesome. so it just allows you to tie a callback to an External that'll run when gc'd?
17:53:58  <indutny>not when gc'd
17:54:00  <MI6>joyent/node: isaacs v0.10 * 164d5b3 : tls: Destroy socket when encrypted side closes The v0.8 Stream.pipe() me - http://git.io/8moh0w
17:54:02  <indutny>when its retained
17:54:17  <indutny>that's a big difference :)
17:54:21  <indutny>but its how GC actually works
17:54:32  <indutny>well, I'll probably rewrite it a bit
17:54:39  <indutny>and use weak references instead...
17:54:53  <bradleymeck>well, modern GC works by moving retained, old as hell GC ...
17:54:54  <indutny>but it would be good if you'll tell me if its useful for you in this form
17:55:05  <indutny>bradleymeck: not always
17:55:22  <indutny>anyway
17:55:36  <bradleymeck>always is a hard word to keep to in computing
17:55:36  <indutny>piscisaureus_: "Honestly if her solution for improving TLS performance wasn’t so damn elegant, I would be pretty upset about that whole JavaSwift thing"
17:55:40  <indutny>oh you naughty boy
17:56:31  <isaacs>indutny: context?
17:56:38  <indutny>http://blog.strongloop.com/taylor-swift-now-dating-node-js/
17:56:46  <isaacs>oh, i see.
17:56:51  <isaacs>ugh. fucking april fools day.
17:57:15  <tjfontaine>heh
17:59:26  * inolenpart
18:00:17  <bnoordhuis>fwiw, i wasn't involved
18:00:33  <tjfontaine>bnoordhuis: I can close https://github.com/joyent/node/issues/2301 right?
18:01:06  <indutny>bnoordhuis: so that's bert's solely guilty?
18:01:18  <bnoordhuis>tjfontaine: yep, it's implemented
18:01:37  <tjfontaine>thanks
18:01:52  <indutny>bnoordhuis: btw, about tls
18:01:57  <bnoordhuis>indutny: our ghost writer, actually
18:02:08  <indutny>can you show me some benchmarks that you performed recently?
18:02:16  <indutny>I wonder how you benchmarked eNULL cipher...
18:02:38  <trevnorris>indutny: sorry, what specifically do you mean by "retained". don't want to rely on what I _think_ it means.
18:02:47  <indutny>trevnorris: object is alive after gc
18:02:54  <trevnorris>got it.
18:03:02  <bnoordhuis>indutny: there's only one aNULL+eNULL cipher actually
18:03:10  <bnoordhuis>or at least, only one that openssl supports
18:03:15  <indutny>bnoordhuis: I believe you
18:03:23  <indutny>but... nor ab, nor siege doesn't support it AFAIK
18:03:28  <indutny>h
18:03:30  <indutny>ab supports it
18:03:31  <indutny>ok
18:03:59  <tjfontaine>this isn't valid anymore either is it? https://github.com/joyent/node/issues/2586
18:04:22  <trevnorris>indutny: i like it. this would remove the need to persist/makeweak.
18:04:38  <indutny>yes, but is it *that slow* now?
18:04:45  <bnoordhuis>tjfontaine: well... maybe leave it open as a rainy day project?
18:05:22  <bnoordhuis>it's not something i care about personally but maybe others do
18:05:35  <indutny>well, I like it
18:05:36  <tjfontaine>I thought we were already doing this?
18:05:43  <indutny>tjfontaine: no, not really
18:05:46  <indutny>fs.open() returns fd
18:05:50  <tjfontaine>oh right
18:06:26  <bnoordhuis>leaking file descriptors was something of an issue in cloud9
18:06:39  <bnoordhuis>so it's probably something other people run into as well
18:07:11  <indutny>that's right
18:07:17  <indutny>I might look into it
18:07:19  <indutny>on this week
18:07:24  <trevnorris>indutny: well, once I figured out the correct syntax it wasn't *that slow* to instantiate. would this be any less stressful on gc?
18:07:35  <indutny>trevnorris: I'm not sure
18:07:45  <indutny>suppose overhead is not that big
18:08:06  * dominictarrquit (Quit: dominictarr)
18:09:47  <indutny>bnoordhuis: so, can you please help me with getting this benchmark runnig?
18:09:54  <indutny>s/runnig/running/
18:09:59  <trevnorris>well i like how straight forward it is. can the external memory be tracked by heapdump?
18:10:05  <indutny>ab -Z eNULL doesn't seem to be working
18:10:07  <bnoordhuis>indutny: sure. with what in particular?
18:10:21  <indutny>bnoordhuis: what cipher have you used?
18:10:40  <bnoordhuis> openssl ciphers aNULL+eNULL
18:10:40  <bnoordhuis>AECDH-NULL-SHA
18:10:45  <bnoordhuis>^ indutny
18:10:55  <MI6>nodejs-v0.10: #97 UNSTABLE osx-x64 (1/570) windows-x64 (4/570) smartos-x64 (1/570) smartos-ia32 (1/570) windows-ia32 (4/570) http://jenkins.nodejs.org/job/nodejs-v0.10/97/
18:11:08  <bnoordhuis>if that doesn't work, use e.g. NULL-MD5
18:11:14  * ryahquit (Quit: leaving)
18:11:16  <bnoordhuis>that's eNULL, just not aNULL
18:12:25  <indutny>SSL handshake failed
18:12:25  <indutny>lets try siege
18:12:44  <bnoordhuis>make sure your server allows the cipher
18:12:49  <bnoordhuis>they're disabled by default
18:13:06  <indutny> ciphers: 'NULL-MD5'
18:13:12  <indutny>I guess this should do it
18:13:15  * qmx|lunchchanged nick to qmx
18:13:19  <indutny>that's `https` server option
18:13:28  * kazuponquit (Remote host closed the connection)
18:13:30  <bnoordhuis>let me check what i have
18:13:56  <indutny>sure
18:14:08  <bnoordhuis>sigh. need to walk upstairs and turn on the computer
18:15:03  * benoitcquit (Excess Flood)
18:17:31  <indutny>bnoordhuis: so you're not going there? :)
18:18:20  <piscisaureus_>isaacs: the docs for v0.11.0 aren't on the website?
18:20:11  <bnoordhuis>indutny: already back
18:20:42  <bnoordhuis>i think i hacked ab to allow AECDH-NULL-SHA but NULL-MD5 should work
18:20:46  <bnoordhuis>let me test it
18:20:53  <tjfontaine>piscisaureus_: https://github.com/joyent/node/issues/2738 can be closed since uv_poll was added?
18:21:06  <indutny>bnoordhuis: its quite interesting
18:21:07  <indutny>but
18:21:20  <bnoordhuis>yeah, wfm
18:21:20  <indutny>AECDH-NULL-SHA
18:21:24  <indutny>err
18:21:25  <indutny> SSL_CTX_set_cipher_list(ctx_, "AECDH-NULL-SHA");
18:21:32  <indutny>this doesn't seem to work in tlsnappy
18:21:39  * benoitcjoined
18:22:11  <piscisaureus_>tjfontaine: the HandleWrap issue should stay open
18:22:17  <tjfontaine>nod
18:23:19  <bnoordhuis>indutny: it's a diffie-hellman cipher
18:23:28  <bnoordhuis>i.e. not a regular ssl cipher
18:23:35  <indutny>gosh
18:23:49  <bnoordhuis>don't worry about it, NULL-MD5 is good enough for benchmarking
18:24:38  <indutny>well, it doesn't work either :)
18:24:43  <bnoordhuis>no? how come?
18:25:10  <indutny>well, might be I shouldn't add RSA cert+key to it
18:25:26  <indutny>but client still chooses this:
18:25:27  <indutny>New, TLSv1/SSLv3, Cipher is RC4-SHA
18:26:30  <isaacs>piscisaureus_: uploading now
18:26:42  <bnoordhuis>indutny: i don't know, wfm
18:26:44  <indutny>bnoordhuis: removing certificates doesn't help either
18:26:54  <indutny>ok, going to debug it a little
18:27:01  <bnoordhuis>indutny: what does `curl -kv --ciphers NULL-MD5 <url>` print?
18:27:34  <indutny>curl: (35) error:14077410:SSL routines:SSL23_GET_SERVER_HELLO:sslv3 alert handshake failure
18:27:40  <indutny>I've only one thing to blame
18:27:41  <indutny>osx
18:27:53  <bnoordhuis>hah, quite right :)
18:28:08  <bnoordhuis>i never realized this but it's not possible to start a tls/https server in node with no key/cert
18:28:25  <bnoordhuis>probably no one cares about that though
18:28:28  <indutny>heh
18:28:33  <indutny>its better this way
18:28:42  <indutny>otherwise people will try to start "normal" server without them
18:28:54  <tjfontaine>I hit an issue over the weekend that I wasn't able to see an error for the fact I had a mismatched cert/key pair
18:28:54  <bnoordhuis>yeah. less room for mistakes
18:29:11  <isaacs>https://twitter.com/nodejs/status/318792106879623170
18:29:24  <bnoordhuis>tjfontaine: oh? that sounds like a bug
18:30:04  <tjfontaine>bnoordhuis: ya, I didn't spend too much time on it, when I realized I was an idiot I moved on I didn't look to see if there was an error somewhere that I simply wasn't seeing
18:30:24  <tjfontaine>but generally emit('error') without a listener throws, right?
18:30:31  <indutny>yeah
18:30:52  <tjfontaine>so, unless it was supposed to go to another cb then it's probably a bug
18:34:27  <tjfontaine>ok, it does throw in clientError but it's not particularly helpful, Error: 140735180325248:error:1408A0C1:SSL routines:SSL3_GET_CLIENT_HELLO:no shared cipher:../deps/openssl/openssl/ssl/s3_srvr.c:1355:
18:34:46  <tjfontaine>we should be able to at least know that there was a key/cert failure
18:35:19  <bnoordhuis>ah right
18:35:32  <bnoordhuis>the error message is factual and correct as far as these things go on
18:35:39  <bnoordhuis>but not very helpful, no
18:35:43  <tjfontaine>right
18:36:07  <txdv>promises? 1 april?
18:36:18  <bnoordhuis>s/go on/go/
18:36:31  <indutny>txdv: no, its just time for them
18:36:38  <bnoordhuis>tjfontaine: open an issue, it's probably something we can fix
18:36:40  <indutny>people are rejecting node without them
18:36:44  <tjfontaine>bnoordhuis: ok
18:37:13  <bnoordhuis>promises isn't taking it far enough. fibers!
18:37:26  <indutny>bnoordhuis: heretic!
18:40:22  <txdv>so the api of a stable version changes?
18:40:59  <indutny>txdv: yeah, a bit
18:41:33  <txdv>the shitstorm is comming
18:42:01  * hzquit
18:42:27  * hzjoined
18:43:51  * kazuponjoined
18:43:52  * benoitcquit (Excess Flood)
18:44:10  <indutny>bnoordhuis: my bad
18:44:16  <indutny>bnoordhuis: I was overwriting ciphers from js
18:44:17  <indutny>:(
18:45:39  <indutny>works fine now
18:47:39  * benoitcjoined
18:49:48  <indutny>bnoordhuis: what have you benchmarked?
18:49:51  <indutny>bnoordhuis: big responses?
18:49:53  <indutny>or just rps?
18:49:56  <indutny>i.e.
18:49:59  <indutny>throughput or rps?
18:50:06  * mikealquit (Quit: Leaving.)
18:52:26  * kazuponquit (Ping timeout: 256 seconds)
18:53:24  <bnoordhuis>indutny: neither. i used it for profiling
18:53:37  <indutny>well, what you was profiling?
18:53:44  <bnoordhuis>well, node
18:53:59  <indutny>well, that's not descriptive
18:53:59  <bnoordhuis>in particular, checking where it spends most of its time
18:54:09  <indutny>spends most time when doing what?
18:54:21  <bnoordhuis>...
18:54:57  <indutny>you must agree this is an important question
18:55:03  <indutny>with tls
18:55:06  * c4miloquit (Remote host closed the connection)
18:55:12  <indutny>you're either benchmarking handshakes, or throughput
18:55:19  <bnoordhuis>oh, is that what you mean
18:55:29  <indutny>I believe it is
18:55:33  * c4milojoined
18:55:45  <bnoordhuis>both, in that case
18:55:58  <indutny>ok
18:56:08  <indutny>so I can't see where it spends a lot of time allocating data
18:56:12  <indutny>its still bn_mul_words for me
18:56:13  <indutny>mostly
18:56:21  <bnoordhuis>wut?
18:56:24  <indutny>for both tlsnappy and node.js
18:56:29  <bnoordhuis>are you sure you're using a null cipher?
18:56:30  <indutny>bnoordhuis: yeah
18:56:50  <indutny>SSL/TLS Protocol: TLSv1/SSLv3,NULL-MD5,1024,0
18:57:19  <indutny>i belive it still does DH exchange
18:57:30  <indutny>aah
18:57:34  <indutny>that's private key decrypt
18:57:44  <indutny>well
18:57:46  <indutny>anyway
18:57:54  <indutny>next to it
18:57:56  <indutny>is write() call
18:58:20  <indutny>and malloc is only about 1.8% of total execution
18:58:24  <indutny>at least for tlsnappy
18:59:06  <bnoordhuis>i'll tell you tomorrow, i turned my upstair's pc off again :/
18:59:08  <indutny>seems to be not affecting it
18:59:10  <indutny>haha
19:00:06  * c4miloquit (Ping timeout: 256 seconds)
19:00:11  <bnoordhuis>i guess i can run some quick benchmarks on this MBA
19:00:26  <bnoordhuis>but well, benchmarks and os x...
19:00:30  <indutny>meh
19:00:35  <indutny>one thread should work pretty well :)
19:00:49  <indutny>anyway
19:00:52  <indutny>you should get slower allocations
19:01:02  <indutny>and they should be more noticable ;)
19:01:31  <indutny>perhaps, I'm doing something wrong
19:01:35  <indutny>but I can't tell what
19:04:53  <bnoordhuis>god, dtrace-on-os-x and ustack()
19:05:08  <bnoordhuis>dtrace: error on enabled probe ID 1 (ID 49713: profile:::profile-1001): invalid address (0x7fff5fbf5db0) in action #2
19:05:12  <bnoordhuis>i get like a million of these
19:05:27  <indutny>haha
19:05:30  <indutny>use instruments
19:05:37  <indutny>Instruments.app
19:08:36  <bnoordhuis>i have no idea how instruments works, truth be told
19:08:47  <bnoordhuis>it's so... visual
19:08:52  <indutny>yeah, I know
19:09:02  * xaqjoined
19:09:08  <indutny>just choose time profiler
19:09:21  <indutny>Attach to node process
19:09:23  <indutny>and click record
19:10:01  * indexzeroquit (Quit: indexzero)
19:13:01  <bnoordhuis>i think i broke my script. i'm getting 'SSL handshake failed' errors
19:13:27  <bnoordhuis>then again, this _is_ os x
19:14:06  <bnoordhuis>curl doesn't work either :-/
19:14:09  * indexzerojoined
19:14:26  <trevnorris>would you two say most buffer allocations are > or < 256 bytes?
19:16:53  <bnoordhuis>um... ssertion failed: (handle_->Get(String::New("error"))->BooleanValue() == false), function ClearError, file ../../src/node_crypto.cc, line 985.
19:17:09  <bnoordhuis>trevnorris: what buffer allocations?
19:18:13  <trevnorris>heh, that was open-ended. i mean for how net/http/tls allocate buffers for sending/receiving data.
19:19:00  <bnoordhuis>trevnorris: are we talking about `new Buffer(...)` in js land or Buffer::New(...) in c++ land
19:19:22  <bnoordhuis>aside, i have grave doubts about dtrace's accuracy as a profiler
19:19:28  <trevnorris>more new Buffer(). my allocator already kicks the shiz outta Buffer::New.
19:20:22  <bnoordhuis>trevnorris: right. hard to say, really
19:20:31  * mikeal1joined
19:20:37  <bnoordhuis>it probably depends on the workload
19:22:23  <trevnorris>bnoordhuis: true. guess it totally depends on how it's being used.
19:22:25  <bnoordhuis>indutny: i'll rerun those benchmarks on linux tomorrow. i don't trust these os x numbers
19:26:14  * paddybyersjoined
19:30:25  <indutny>bnoordhuis: I believe I fixed that error
19:30:29  <indutny>bnoordhuis: what node version are you using?
19:30:39  <indutny>bnoordhuis: and looking forward for results from linux
19:30:42  * mikeal1quit (Ping timeout: 276 seconds)
19:33:47  <eris0xff>isaacs: should I have posted my API idea to the joyent libuv issues instead?
19:39:07  * c4milojoined
19:44:53  * piscisaureus_quit (Ping timeout: 240 seconds)
19:45:38  <tjfontaine>eris0xff: probably, but bnoordhuis should weigh in (#5191)
19:47:49  <bnoordhuis>indutny: master
19:47:54  <indutny>ah
19:47:56  <indutny>merge v0.10
19:48:00  <indutny>or was it merged?
19:48:14  <bnoordhuis>on march 28
19:48:52  <indutny>interesting
19:48:56  <bnoordhuis>eris0xff: that should be a libuv issue
19:49:05  * kazuponjoined
19:52:02  * c4miloquit (Remote host closed the connection)
19:52:17  * benoitcquit (Excess Flood)
19:52:29  * c4milojoined
19:52:59  * c4miloquit (Read error: No route to host)
19:53:22  * c4milojoined
19:53:56  * kazuponquit (Ping timeout: 256 seconds)
19:54:10  <trevnorris>tjfontaine: so my allocator is ~40ns slower than new Buffer(), for buffers < 256 bytes. premature optimization to try fixing that now?
19:55:24  * AvianFluquit (Remote host closed the connection)
19:55:31  <tjfontaine>trevnorris: I guess it depends on if you can isolate why, from what I remember that sounds a lot like the c++ boundary getting in your way
19:56:34  <tjfontaine>I should rebase my changes on top of my stap changes so you can play with the instrumentation stuff
19:56:43  * xaqquit (Remote host closed the connection)
19:57:12  * sgallaghquit (Ping timeout: 264 seconds)
19:57:21  <trevnorris>tjfontaine: yeah. guess I could drop several checks that would bring them on par.
19:57:40  * benoitcjoined
19:57:44  * brsonquit (Remote host closed the connection)
19:57:50  <trevnorris>but just like that saftey net preventing me from a FATAL ERROR
19:58:09  <tjfontaine>bnoordhuis: btw, do you think we could backport the recent dtrace changes to 0.10? shouldn't be too much of a surprise for people on osx since it's just more arguments (that actually work) and makes the fd field actually valid for smartos
20:00:32  <eris0xff>bnoordhuis: Okay. I can refile it under libuv or if you have the capability to move it there you can do it.
20:00:33  * brsonjoined
20:00:37  <tjfontaine>trevnorris: ya, I dunno about the checks, if there's still going to be an outer buffer sanity class and this is just the allocator I would vote in favor as fast as possible with only some bounds checking, maybe comment them out but left for posterity, or hidden behind a NODE_PLEASE_BE_SANE ifdef
20:01:06  <trevnorris>tjfontaine: lol. ok.
20:01:55  <bnoordhuis>tjfontaine: no objections from me
20:02:12  <bnoordhuis>eris0xff: sorry, i / github can't move issues
20:02:28  <bnoordhuis>but just copy/paste it into a new one
20:02:44  <trevnorris>tjfontaine or bnoordhuis: in your opinion, the allocator doesn't need a full class. cool to namespace it like it was (e.g. Smalloc::Alloc)?
20:03:21  <eris0xff>bnoordhuis: okay i've got it
20:04:10  <eris0xff>bnoordhuis: btw: If I'm in an event-loop-less thread, how do I get a "loop" for the uv_last_error?
20:04:16  <tjfontaine>eris0xff: reference the joyent issue in the libuv one, so for people who happen to stumble on it in node know where the new discussion is happening
20:04:25  <eris0xff>Okay.
20:04:34  <tjfontaine>s/joyent/node/
20:04:39  <eris0xff>right
20:04:51  <bnoordhuis>eris0xff: btw, you know sched_yield() is only a hint right?
20:05:16  <tjfontaine>trevnorris: a static method is a static method, for all I care it could be node::Smalloc() :)
20:06:01  <trevnorris>tjfontaine: ok thanks.
20:07:34  <eris0xff>bnoordhuis: interesting, though it seems that in Linux pthread_yield() simply translates to sched_yield(). Lemme check sources for correct semantics.
20:08:10  <bnoordhuis>eris0xff: that's correct. what i mean is that sched_yield() only yields if there's another process running on the same cpu
20:08:16  * sgallaghjoined
20:08:30  <bnoordhuis>iow, if your process is the sole owner of that cpu, it won't yield because there's nothing to yield to
20:08:36  <eris0xff>bnoordhuis: yes. that means that it has the correct semantics.
20:09:03  * paddybyersquit (Ping timeout: 276 seconds)
20:09:21  <eris0xff>bnoordhuis: which brings up another question. I could try-and-try again until I hit a limit, but that would waste considerable cycles on single processor architectures.
20:09:37  <eris0xff>Does libuv have a CPU topology utility to tell me how many cores etc/
20:09:40  * qmxchanged nick to qmx|away
20:09:49  <indutny>bnoordhuis: you still there?
20:09:58  <eris0xff>I know where to get one, but ...
20:10:08  <indutny>bnoordhuis: isaacs: https://github.com/joyent/node/pull/5192
20:13:26  <trevnorris>is there any preference/difference using size_t or uint32_t?
20:14:02  <bnoordhuis>eris0xff: libuv doesn't have anything like that
20:14:24  <bnoordhuis>btw, fun fact about pthread_yield() - there's really no such function
20:14:42  <eris0xff>interesting.
20:14:48  <eris0xff>i thought I've used it before
20:14:51  <eris0xff>(hang on :-)
20:14:52  <bnoordhuis>it was never standarized because it duplicated what sched_yield(0 does
20:15:07  <eris0xff>its def non standard
20:15:11  <bnoordhuis>so the posix guys dropped it in the final draft of the 2001 or 2003 spec
20:15:30  <indutny>trevnorris: its different things
20:15:33  <eris0xff>but does it exist in the user pthreads lib?
20:15:37  <indutny>trevnorris: what do you need?
20:15:39  <bnoordhuis>yeah, most platforms implement it
20:15:42  <indutny>eris0xff: not in every
20:16:02  <bnoordhuis>i just like throwing around useless trivia
20:16:03  <indutny>eris0xff: I believe freebsd has the most fucked posix headers that have ever existed
20:16:10  <eris0xff>hah
20:16:16  <indutny>and it might not include it
20:16:20  <indutny>but bnoordhuis should know it better
20:16:33  <eris0xff>pthread_yield or sched_yield?
20:16:35  <trevnorris>indutny: parameter for length of external array data.
20:16:35  <indutny>bnoordhuis: review review review https://github.com/joyent/node/pull/5192
20:16:40  <indutny>eris0xff: pthread
20:16:43  <eris0xff>ah
20:16:48  <eris0xff>good thing i'm not using it :-)
20:16:57  <indutny>isn't sched_yield linuxism?
20:17:00  <eris0xff>no
20:17:06  * mikealjoined
20:17:07  <eris0xff>its POSIX1 2003-2008
20:17:16  <bnoordhuis>indutny: you know, i haven't been able to reproduce that anymore
20:17:31  <indutny>eris0xff: too fresh to be real :)
20:17:31  <eris0xff>click on helpfully provided link in issue: :-)
20:17:37  <bnoordhuis>and when i went over the xnu source code, i found bits of code that seem like they stop SCM_RIGHTS messages from merging
20:17:53  <indutny>bnoordhuis: ok, so why this issue is open then?
20:18:02  <indutny>:)
20:18:06  <bnoordhuis>because the test is still failing :(
20:18:23  <indutny>well, may be we should try my patch then? ;)
20:18:37  <bnoordhuis>does it fix the test for you?
20:18:37  <indutny>tjfontaine: can you please build this PR?
20:18:43  <indutny>bnoordhuis: it doesn't fail
20:18:44  <indutny>so
20:18:45  <indutny>yes
20:18:52  <bnoordhuis>does it fail without your patch?
20:19:07  <indutny>yes
20:19:14  <bnoordhuis>hrm, interesting
20:19:19  <indutny>indeed
20:19:31  <indutny>it might be some libuv bug
20:19:53  <indutny>but the fact that sending them separately fixes it
20:20:03  <indutny>prooves that its somehow related to merging
20:20:12  <tjfontaine>indutny: I had queued it previously, but apparently faster than what the git repo saw
20:20:52  <indutny>aha :)
20:20:53  <indutny>ok
20:20:55  <indutny>thanks
20:22:25  <tjfontaine>hmm that's still not showing up in the pr refs
20:22:29  * TooTallNatejoined
20:23:14  * sgallaghquit (Remote host closed the connection)
20:24:50  <eris0xff>bnoordhuis: is there provision for "optional" libuv modules?
20:25:37  <tjfontaine>indutny: not sure what to do, github hasn't even sent the email for that pullreq yet, and it's not showing up in the refs yet
20:26:05  <eris0xff>bnoordhuis: and can libuv pull from bsd license?
20:27:00  <bnoordhuis>eris0xff: no and yes
20:27:10  <eris0xff>sad and happy :-)
20:28:07  <eris0xff>okay i'll clone git and you can accept my patches as you see fit
20:31:56  * mikealquit (Quit: Leaving.)
20:35:12  * piscisaureus_joined
20:37:46  <bnoordhuis>god, i just made the newbiest mistake possible
20:37:59  <bnoordhuis>first strlen(buf) and only then fread(buf)...
20:39:27  * mikealjoined
20:39:35  <eris0xff>oops :-)
20:40:03  <tjfontaine>piscisaureus_: child_process.spawn() works with bat and PATHEXT right? iirc that landed for 0.10
20:40:16  <eris0xff>guess you'll have to start using managed code
20:40:29  * benoitcquit (Excess Flood)
20:41:02  <creationix>nah, just write libuv in rust ;)
20:41:12  <creationix>then it will be it's own grandparent or something like that
20:41:47  <eris0xff>lol
20:41:59  <eris0xff>self-hosting event library
20:42:19  <eris0xff>pretty soon you will have removed the machine and your scheme will be complete.
20:47:10  * benoitcjoined
20:49:40  * kazuponjoined
20:49:50  * xaqjoined
20:53:59  <indutny>bnoordhuis: so mind pulling my patch?
20:54:23  * kazuponquit (Ping timeout: 260 seconds)
21:01:52  * rendarquit
21:03:19  <bnoordhuis>indutny: i haven't tried it yet
21:03:26  <bnoordhuis>today's a bank holiday, you know?
21:03:54  <indutny>and?
21:04:00  <indutny>bnoordhuis: you don't have money
21:04:02  <indutny>:)
21:05:50  <bnoordhuis>'tis true, it's all locked up in stock and mutuals
21:05:57  <indutny>obviously
21:06:04  <indutny>banks owes your money
21:06:11  <indutny>err
21:06:14  <indutny>s/owe/own/
21:06:26  <indutny>ok, anyway
21:06:34  <indutny>it seems that you're at reviewal strike again
21:06:43  <indutny>and I'm not going to argue
21:06:46  <indutny>ttyl :)
21:06:47  * indutny&
21:06:48  <LOUDBOT>"DRUNK" IS THE PAST PARTICIPLE
21:07:02  <tjfontaine>LOUDBOT: I think I love you
21:07:02  <LOUDBOT>tjfontaine: WHAT WOULD CARROLL O'CONNER DO?
21:08:45  <bnoordhuis>sleep tight fedor
21:11:52  * benoitcquit (Excess Flood)
21:11:57  * paddybyersjoined
21:15:10  <trevnorris>when were modules like "net" and "fs" included by default in REPL?
21:19:04  <wolfeida_>tjfontaine: I messed around with stap on ubuntu 12.10 and ran into a wall with one last frustrating error
21:19:13  <tjfontaine>oh?
21:19:36  <wolfeida_>https://gist.github.com/wolfeidau/5282942 down the bottom "Error, 'stap_d99f81d30be4ba5ed701ceb812886f06_3723' is not a zombie systemtap module."
21:20:03  <wolfeida_>tjfontaine: That is after hacking thier c code btw
21:20:11  * benoitcjoined
21:20:53  <tjfontaine>wolfeida_: btw, you can just add -vvvv to make your life easier on that verbosity thing
21:21:35  <wolfeida_>tjfontaine: Yeah read an amusing post by brendan which illustrated copying that -vp crap a lot :)
21:21:44  <tjfontaine>:)
21:21:57  <tjfontaine>sudo usermod -G sudo,stapsys,stapdev,stapusr markw <-- that doesn't need an -a with it?
21:22:02  <tjfontaine>not that this matters all that much
21:23:03  * hzquit
21:23:07  <wolfeida_>tjfontaine: Yeah should be append sorry nuked my sudo rites with that, luckily i had another term floating around :)
21:23:15  <tjfontaine>hehe
21:24:17  <wolfeida_>It is still as buggy as hell
21:24:31  <tjfontaine>did you try brendans "fix" for that error?
21:25:16  <tjfontaine>a lot of time when I was testing I ran with --disable-cache as well
21:26:44  <wolfeida_>tjfontaine: yeah removed ~/.systemtap and same error
21:26:52  <wolfeida_>I will try with disable cache
21:27:12  <wolfeida_>I just put that tweak i made to thier sources in a diff
21:27:39  <wolfeida_>They have renamed some functions and the includes for user space probes were broken
21:27:48  <wolfeida_>Like does anyone TEST THIS AT ALL
21:27:50  <tjfontaine>also at one point it just stopped matching probes, but after a reboot it worked :/
21:28:15  <wolfeida_>tjfontaine: Yeah if you list kernel modules loaded i think you can manually unload them?
21:28:25  <tjfontaine>wolfeida_: most people debug with printf, and most things don't ship with symbols -- so no, noone tests it :)
21:28:43  <tjfontaine>wolfeida_: probably, I was just frustrated and hit hte power and was all FUCK THIS
21:28:45  <wolfeida_>tjfontaine: I meant this systemtap stuff :)
21:28:53  <tjfontaine>wolfeida_: me too :)
21:29:01  <wolfeida_>lol
21:29:22  <tjfontaine>add ddebs and download 800mbs worth of symbols so this is useful for you
21:29:23  <tjfontaine>:)
21:30:59  <wolfeida_>tjfontaine: Yeah no go, SO DAMN CLOSE too
21:31:19  <tjfontaine>wolfeida_: do-release-upgrade -d bet it will just work after that :)
21:32:58  <wolfeida_>I will try on Redhat 6 at work today and work back from there
21:33:11  <tjfontaine>best of luck
21:33:18  <wolfeida_>tjfontaine: Cheers
21:33:41  <piscisaureus_>tjfontaine: no .bat doesn
21:33:46  <piscisaureus_>'t work in 0.10
21:33:49  <tjfontaine>hmm
21:34:05  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
21:34:21  <tjfontaine>ah if I specify .bat it works, ok
21:34:42  * piscisaureus_joined
21:40:24  * paddybyersquit (Ping timeout: 264 seconds)
21:42:37  <trevnorris>bnoordhuis: while i'm working on Buffers, and small misc things i should cleanup (e.g. "char *data")?
21:50:47  * kazuponjoined
21:52:01  <trevnorris>TooTallNate: you know if there's a reason why the Buffer:: methods are all accessed like "Buffer::Length()" instead of just "buffer->Length()"?
21:52:31  <trevnorris>or ".Length()" (?)
21:52:45  <trevnorris>(w/o compile errors, I'd really suck at cc)
21:53:44  <tjfontaine>because you're passing in a v8 type you don't have an actual Buffer c++ instance?
21:54:19  <trevnorris>tjfontaine: but it's not uncommon to "Buffer* buffer = ObjectWrap::Unwrap<Buffer>(obj)" beforehand.
21:55:02  <TooTallNate>trevnorris: right, because of SlowBuffers and Buffers basically
21:55:04  <TooTallNate>you can pass in either
21:55:10  <TooTallNate>Buffer instances or V8 objects
21:55:26  <tjfontaine>regardless of what you have, you have the same interface :)
21:55:36  <TooTallNate>trevnorris: i think the API was like you're asking for pre v0.3 of node :)
21:55:39  <TooTallNate>a loonnngggg time ago :p
21:55:46  * kazuponquit (Ping timeout: 256 seconds)
21:55:50  <tjfontaine>ya, the header talks about the upgrade path
21:55:54  <tjfontaine>return Buffer::Length(b->handle_);
21:56:14  <trevnorris>but that was for when they went to Buffers/SlowBuffers. with the new allocator they'd be the same.
21:56:37  <tjfontaine>trevnorris: perhaps but as far as an upgrade path you'd have to make sure it still works
21:56:47  <tjfontaine>even if it does something different as far as implementation
21:57:16  <trevnorris>tjfontaine: oh yeah. the api is going to be exactly the same, except for a few necessary additions.
21:57:35  <bnoordhuis>trevnorris: re cleanups, sure, but keep them in separate commits unless you're changing that line anyway
21:57:59  <trevnorris>tjfontaine: just want to skip the step of "GetIndexedPropertiesExternalArrayData" and instead "return data_"
21:58:27  <tjfontaine>nod
21:59:13  <trevnorris>awesome. thanks guys. bbiab
21:59:15  * trevnorrisquit (Quit: Leaving)
21:59:27  <tjfontaine>I will teach him to idle yet
22:02:55  * mikealquit (Quit: Leaving.)
22:03:07  * xaqquit (Remote host closed the connection)
22:05:02  * wolfeida_quit (Remote host closed the connection)
22:07:58  <isaacs>tjfontaine: whoa! https://github.com/joyent/node/pull/5186 http://jenkins.nodejs.org/job/node-pullrequest/114/
22:08:02  <isaacs>tjfontaine: lots of failures there
22:09:13  <tjfontaine>I added windows to the pullreq's since I missed that from the etw stuff before, so those windows ones aren't surprising
22:09:36  <isaacs>oh, that's master...
22:09:40  <isaacs>maybe that's it?
22:09:45  <tjfontaine>ya
22:10:00  <tjfontaine>the smartos failures I've changed the resolvers since to googles to see if that clears that up, but I'm not sure why we're seeing intermitent failures on those
22:10:00  <isaacs>clearly this belongs in 0.10. ok
22:10:02  <isaacs>testing now
22:12:44  * paddybyersjoined
22:13:45  * c4miloquit (Remote host closed the connection)
22:14:12  * c4milojoined
22:15:45  * xaqjoined
22:16:02  * indexzeroquit (Quit: indexzero)
22:16:04  <TooTallNate>piscisaureus_: nice!
22:16:06  <TooTallNate>re: gypified
22:16:20  <piscisaureus_>TooTallNate: I guessed as much. Now it's up to you though :-p
22:16:37  <TooTallNate>yup, i got like probably 10+ libs to add
22:16:37  <piscisaureus_>Maybe we should only put repos there that we really want to keep up to date ...
22:17:02  <TooTallNate>they're all ones that i'm using in node addons at this point
22:17:08  <piscisaureus_>TooTallNate: feel free to delete repos, too
22:18:33  * c4miloquit (Ping timeout: 240 seconds)
22:26:52  * indexzerojoined
22:27:15  * benoitcquit (Excess Flood)
22:29:37  * indexzeroquit (Client Quit)
22:30:41  * benoitcjoined
22:31:40  * paddybyersquit (Ping timeout: 272 seconds)
22:33:56  <tjfontaine>I feel like I missed something
22:35:46  <TooTallNate>tjfontaine: https://github.com/gypified
22:35:53  <TooTallNate>i'm gonna add libffi and some others to that
22:36:13  <tjfontaine>ah
22:48:17  <MI6>joyent/node: Andrew Hart v0.10 * db8ce89 : test: test intended code-paths The tests did not agree with the test com - http://git.io/BvHR2A
22:51:22  * kazuponjoined
22:52:42  * wolfeidaujoined
22:54:17  * trevnorrisjoined
22:55:19  * paddybyersjoined
22:56:06  * kazuponquit (Ping timeout: 256 seconds)
22:59:11  <tjfontaine>oops, looks like > and not >=
22:59:18  <tjfontaine>er anyway
23:01:15  * xaqquit (Remote host closed the connection)
23:01:27  <trevnorris>tjfontaine: promise i'll setup a persistent client once my initial buffer changes are accepted. ;-)
23:01:39  <tjfontaine>haha
23:03:38  * bradleymeckquit (Quit: bradleymeck)
23:03:54  * paddybyersquit (Ping timeout: 276 seconds)
23:06:02  <MI6>nodejs-v0.10: #98 UNSTABLE windows-x64 (5/570) osx-ia32 (1/570) smartos-ia32 (1/570) windows-ia32 (4/570) linux-ia32 (1/570) http://jenkins.nodejs.org/job/nodejs-v0.10/98/