00:00:15  <ryah>pquerna: you have to watch the nextgen crytpo wwdc11 video
00:00:18  <ryah>if you haven't already
00:01:34  <CIA-53>libuv: Ben Noordhuis * rf7f518a / src/unix/process.c :
00:01:34  <CIA-53>libuv: sunos: fix uv_spawn()
00:01:34  <CIA-53>libuv: SunOS does not set POLLHUP when the read end of a pipe is closed. - http://git.io/LGojxg
00:03:20  <ryah>we're going to have to start building openssl into node for OSX too
00:04:19  <bnoordhuis>because of that lion deprecation thing?
00:04:29  <ryah>yeah
00:05:40  <CIA-53>node: Ryan Dahl * r75cfc77 / (deps/uv/src/unix/process.c deps/uv/uv.gyp): Upgrade libuv to f7f518a - http://git.io/XFIB6A
00:12:04  * sh1mmerjoined
00:19:13  <bnoordhuis>thanks for putting up the slides, ryah
00:19:52  <bnoordhuis>not really sold on that commoncrypto / cf stuff but oh well
00:20:09  <CIA-53>node: Ben Noordhuis * r9bbca99 / deps/v8/src/platform-solaris.cc : v8: implement VirtualMemory class on SunOS - http://git.io/rGQNcQ
00:23:57  <ryah>let's try this again
00:23:58  <ryah>http://nodejs.org/dist/v0.5.9/node-v0.5.9rc2.tar.gz
00:25:20  * bnoordhuisdownloads
00:26:07  <bnoordhuis>you've been doing the releases lately, ryah
00:26:33  <bnoordhuis>wasn't the plan to switch around?
00:28:00  <ryah>yeah - so much pain though
00:28:02  <ryah>next one
00:31:04  <ryah>im building .exe i'll upload it in a minutes
00:33:19  <bnoordhuis>great minds - i'm trying to build it in vs 2010
00:33:44  <bnoordhuis>i've got to hand it to MS, visual studio is a pretty awesome IDE
00:38:18  * erickt_quit (Quit: erickt_)
00:39:07  <ryah>groan. win build fail.
00:40:53  <bnoordhuis>ryah: isnan / isfinite?
00:41:22  <ryah>no - linking to v8_snapshot
00:41:40  <bnoordhuis>right, then i'm doing something wrong
00:42:07  <bnoordhuis>i get lots of these:
00:42:08  <bnoordhuis>z:\src\node\deps\v8\src\conversions-inl.h(84): error C3861: 'isnan': identifier not found
00:42:08  <bnoordhuis>z:\src\node\deps\v8\src\conversions-inl.h(85): error C3861: 'isfinite': identifier not found
00:42:16  * piscisaureus_quit (Quit: Leaving.)
00:42:58  <bnoordhuis># Fatal error in /home/bnoordhuis/src/nodejs/node-v0.5.9/deps/v8/src/incremental-marking.cc, line 713
00:42:59  <bnoordhuis># CHECK(Marking::IsGrey(Marking::MarkBitFrom(obj))) failed
00:43:01  <ryah>hm
00:43:03  <bnoordhuis>haven't seen that one before
00:44:14  * piscisaureusjoined
00:54:43  <ryah>gyp headache
00:55:41  <bnoordhuis>[17:10|% 100|+ 561|- 17]: Done <- linux
00:57:09  <ryah>does ./node_g test/simple/test-stream-pipe-multi.js break for you?
00:57:50  <ryah>i think we might need to revert the v8 upgrade.
00:57:51  <bnoordhuis>ryah: yes and hard
00:57:54  <bnoordhuis>yes
00:59:08  * piscisaureusquit (Read error: Connection reset by peer)
00:59:09  * piscisaureus_joined
00:59:15  * piscisaureus_changed nick to piscisaureus
00:59:36  <ryah>too bad
00:59:54  <CIA-53>node: Ryan Dahl * r3b1d656 / (282 files in 23 dirs):
00:59:54  <CIA-53>node: Revert "Upgrade V8 to 3.6.6"
00:59:54  <CIA-53>node: Not stable enough.
00:59:54  <CIA-53>node: - Windows snapshot linking broken
00:59:54  <CIA-53>node: - Linux crash on ./node_g test/simple/test-stream-pipe-multi.js
00:59:55  <CIA-53>node: This reverts commit 56e6952e639ba1557a5b22333788583e9e39fa29. - http://git.io/sFDgAw
01:02:59  <ryah>https://twitter.com/#!/kisshotch/status/123561351137329152 <-- people just tweet random shit, huh?
01:06:05  <bnoordhuis>haha wtf?
01:06:23  <bnoordhuis>retweeted even
01:08:32  <ryah>http://nodejs.org/dist/v0.5.9/node-v0.5.9rc3.tar.gz
01:08:45  <ryah>retweeted by the CTO of joyent even :P
01:09:22  * AvianFlu_joined
01:11:20  <bnoordhuis>okay, last one
01:11:28  <bnoordhuis>after rc3 i'm off to bed
01:11:35  * AvianFluquit (Ping timeout: 260 seconds)
01:13:26  * AvianFlu_changed nick to AvianFlu
01:15:35  * sh1mmerquit (Quit: sh1mmer)
01:16:04  <ryah>mother fucker
01:16:16  <ryah>rc3 is broken on solaris
01:16:19  <ryah>god damn it.
01:17:50  * AvianFlu_joined
01:18:41  <CIA-53>node: Ryan Dahl * r07703fa / deps/v8/src/platform-solaris.cc :
01:18:41  <CIA-53>node: Revert "v8: implement VirtualMemory class on SunOS"
01:18:41  <CIA-53>node: This reverts commit 9bbca99107652906a060679ee95bf1ad7381cbb5. - http://git.io/dOSuSQ
01:19:20  <bnoordhuis>okay - but rc4 is really the last one...
01:20:56  * AvianFluquit (Ping timeout: 260 seconds)
01:22:40  <ryah>http://nodejs.org/dist/v0.5.9/node-v0.5.9rc4.tar.gz
01:24:43  * ericktquit (Quit: erickt)
01:26:03  * AvianFlu_changed nick to AvianFlu
01:38:47  * brsonquit (Ping timeout: 252 seconds)
01:51:42  <ryah>indutny:
01:51:42  <ryah>=== release test-debugger-repl ===
01:51:43  <ryah>Path: simple/test-debugger-repl
01:51:43  <ryah>AssertionError: /break in .*:3/
01:51:43  <ryah>Command: out/Release/node /home/ryan/node-v0.5.9/test/simple/test-debugger-repl.js
01:54:32  <bnoordhuis>yeah, i have that too
01:54:35  <bnoordhuis>[16:59|% 100|+ 563|- 15]: Done <- linux
01:55:17  <ryah>i think we have a winner
01:56:30  <bnoordhuis>[15:01|% 100|+ 554|- 24]: Done <- sunos
01:56:56  <ryah>we got to do some work on ths write-callbacks still
01:57:06  <ryah>test/pummel/test-net-write-callbacks.js
01:57:17  <bnoordhuis>yep, times out
01:59:08  * bradleymeckjoined
02:00:47  <bnoordhuis>ryah: don't want to desert you but i need to get some sleep
02:02:01  <ryah>bnoordhuis: please do - just making the announcement
02:02:05  <ryah>bnoordhuis: see you tomororw
02:02:09  <bnoordhuis>sleep tight
02:02:19  * bnoordhuisquit (Quit: Leaving)
02:08:05  * ericktjoined
02:09:34  <CIA-53>node: Ryan Dahl * r3bd9b08 / (5 files in 3 dirs): Bump version to v0.5.9 - http://git.io/wiAq_Q
02:09:34  <CIA-53>node: Ryan Dahl * r24dabec / src/node_version.h : now working on v0.5.10 - http://git.io/dHRfsA
02:21:26  <igorzi_>ryah: piscisaureus: https://gist.github.com/1277039
02:23:06  <igorzi_>basically, with the way we typically run http-simple (with many concurrent clients), doing only 1 pending accept slows it down by ~25%
02:25:29  <igorzi_>with only 32 concurrent clients, doing many pending accepts doesn't scale very well (because all connections are going to the same process)
02:26:55  * bradleymeckquit (Ping timeout: 248 seconds)
02:29:34  <igorzi_>ryah: piscisaureus: so, just going with either approach (statically hardcoding the number of pending accepts) will slow down some benchmarks significantly
03:34:07  * bradleymeckjoined
03:55:30  * bradleymeckquit (Quit: Leaving)
04:07:11  * piscisaureusquit (Ping timeout: 248 seconds)
04:36:59  * isaacsquit (Quit: isaacs)
04:42:11  <indutny>ryah: hi
04:42:19  <indutny>ryah: going to take a look at debugger test
04:49:42  * ericktquit (Quit: erickt)
04:59:12  * AvianFluquit (Quit: Leaving)
05:13:06  * DrPizzaquit (*.net *.split)
05:14:07  * DrPizzajoined
05:14:37  * DrPizzaquit (Excess Flood)
05:14:44  * DrPizzajoined
06:02:47  * AvianFlujoined
06:33:51  * mralephjoined
07:15:22  <ryah>igorzi_: dang :/
07:24:46  * mralephquit (Quit: Leaving.)
08:45:58  * mralephjoined
08:51:56  * piscisaureusjoined
09:11:22  * piscisaureusquit (Ping timeout: 244 seconds)
10:03:39  * mikealquit (Quit: Leaving.)
10:04:20  * mikealjoined
10:08:34  * Casanjoined
10:22:34  <mraleph>ryah: why u no report assertion failures in V8 it would help to stabilize trunk
11:52:25  * mikealquit (Quit: Leaving.)
11:52:53  * mikealjoined
12:11:49  * mikealquit (Quit: Leaving.)
12:19:25  * mikealjoined
12:31:13  * mikealquit (Ping timeout: 245 seconds)
12:32:27  * mikealjoined
12:35:30  * piscisaureusjoined
12:48:57  * mikealquit (Quit: Leaving.)
13:15:23  * bnoordhuisjoined
13:15:32  * mikealjoined
13:24:02  * mikealquit (Quit: Leaving.)
13:30:29  <bnoordhuis>piscisaureus: how was the talk?
13:30:44  <piscisaureus>acceptab;e
13:30:48  <piscisaureus>but the beamer didn't work
13:31:00  <bnoordhuis>hah, that doesn't help
13:31:01  <piscisaureus>which was really annoying
13:31:18  <bnoordhuis>they didn't have a spare?
13:31:30  <bnoordhuis>first rule of conventions, always have a spare beamer
13:31:33  <piscisaureus>it ate 10 minutes off my talk and I had to reboot mac os because the technician wanted to do funky stuff
13:31:40  <piscisaureus>so I couldn't do all of my demos
13:32:09  <piscisaureus>and the demo backup didn't work either because my no.de instance was broken again
13:32:14  <bnoordhuis>aww
13:32:21  <bnoordhuis>one thing after another
13:32:23  <piscisaureus>apart from that it went fine
13:32:46  <bnoordhuis>how many people were there?
13:39:07  * piscisaureusquit (Ping timeout: 276 seconds)
13:43:48  * ericktjoined
14:05:28  * ericktquit (Quit: erickt)
14:06:25  * bradleymeckjoined
14:21:10  * piscisaureusjoined
14:24:25  * ericktjoined
14:24:27  * DrPizzaquit (Excess Flood)
14:24:33  * DrPizzajoined
14:34:47  * pieternjoined
14:53:10  <CIA-53>libuv: Ben Noordhuis * rb3a003f / (test/test-list.h test/test-spawn.c): test: uv_process_kill(proc, 0) should not kill the process - http://git.io/u5tjkQ
14:53:11  <CIA-53>libuv: Ben Noordhuis * r69bd8c3 / src/unix/kqueue.c : unix: stub kqueue support functions on old platforms - http://git.io/eq8C1w
14:53:27  <bnoordhuis>ffs
14:54:11  <CIA-53>libuv: Ben Noordhuis * r47d22ac / src/unix/kqueue.c : unix: stub kqueue support functions on old platforms - http://git.io/ZpsVog
14:56:56  * pieternquit (Quit: pietern)
14:56:59  * mikealjoined
14:57:40  <CIA-53>node: Ben Noordhuis * rc215262 / deps/uv/src/unix/kqueue.c : uv: upgrade to 47d22ac - http://git.io/SqgcNg
15:05:34  * ericktquit (Quit: erickt)
15:14:09  <bnoordhuis>piscisaureus or igorzi_: i can't get v8 to compile with vs 2010 express unless i apply this patch -> https://gist.github.com/2a60fd155b95e2cfadc6
15:14:15  <bnoordhuis>is that reasonable / expected?
15:16:26  <bnoordhuis>hrm, it's probably a gyp thing - the sconstruct file defines WIN32, the gyp file does not
15:21:49  <rmustacc>bnoordhuis: Looks like you did the SunOS stuff for the new v8. Thanks! I was afraid I was going to have to do it.
15:22:35  <bnoordhuis>rmustacc: my pleasure, it was mostly a matter of copy/pasting stuff from platform-linux.cc
15:23:31  <CIA-53>node: Ben Noordhuis * r6cb1d26 / deps/v8/build/common.gypi : v8: fix gyp build on windows - http://git.io/dJ7HGw
15:29:55  * mikealquit (Quit: Leaving.)
15:38:38  * isaacsjoined
15:43:47  <bnoordhuis>to whom it may concern: http://code.google.com/p/v8/issues/detail?id=1760 and http://code.google.com/p/v8/issues/detail?id=1761
15:57:08  * brsonjoined
16:01:31  * casper_joined
16:01:59  * piscisaureusquit (Quit: ~ Trillian Astra - www.trillian.im ~)
16:02:41  <bnoordhuis>oh, and this: http://code.google.com/p/gyp/issues/detail?id=216#c2
16:04:45  * Casanquit (Ping timeout: 258 seconds)
16:11:27  * bradleymeckquit (Ping timeout: 248 seconds)
16:15:02  <bnoordhuis>isaacs: ping
16:15:15  <isaacs>hi
16:15:30  <bnoordhuis>hey
16:15:36  <bnoordhuis>question about the zlib bindings
16:15:41  <isaacs>k
16:15:53  <bnoordhuis>what is stopping the in/out buffer from being garbage collected?
16:16:10  <isaacs>i believe it's attached to the reqwrap object
16:16:34  <bnoordhuis>to wit: https://github.com/joyent/node/blob/6cb1d26924f5dc5416bdb8e04031ad1b41bbc7b5/src/node_zlib.cc#L105
16:16:41  <bnoordhuis>but that doesn't stop v8 from gc'ing, right?
16:16:54  <bnoordhuis>with 'it' i mean the js buffer object
16:17:22  <isaacs>hm.... oh, i think that's just managed from js
16:17:50  <bnoordhuis>yes, i saw that in zlib.js
16:17:54  <isaacs>so the js layer does binding.write(flush, inp, in_off, in_len, etc
16:17:58  <bnoordhuis>it's pushed into a queue
16:18:01  <isaacs>and manages that buffer
16:18:07  <isaacs>as long as it's got a ref to it, it won't be gc'ed
16:18:08  <bnoordhuis>but if the zlib object is gc'ed, then so is the buffer
16:18:23  <isaacs>if the zlib object is gc'ed you can't use it anyway :0
16:18:27  <isaacs>it means you dropped the reference to it
16:18:30  * casper__joined
16:19:14  <bnoordhuis>all that is true
16:19:54  <bnoordhuis>but the worker in the thread pool may still try to write to the memory, right?
16:20:03  <bnoordhuis>i think we have a scoping issue here
16:20:28  <isaacs>you mean if you write to the output buffer after asking zlib to do something?
16:20:38  <isaacs>or write to the input buffer after handing it off to zlib?
16:21:13  <isaacs>once you pass a buffer to a binding, it's not your buffer any more. known issue. you can do the same thing with fs.read.
16:22:05  <bnoordhuis>no, that's not what i mean
16:22:12  <bnoordhuis>i pass in a buffer
16:22:23  * casper_quit (Ping timeout: 258 seconds)
16:22:26  <bnoordhuis>then drop the reference to the zlib object
16:22:36  <bnoordhuis>(and the buffer)
16:22:45  <bnoordhuis>the zlib and buffer objects get gc'ed
16:23:13  <bnoordhuis>2 milliseconds later the thread pool tries to write to the memory the buffer pointed to
16:27:39  * piscisaureusjoined
16:28:41  <CIA-53>libuv: Ben Noordhuis * r6936d60 / config-unix.mk :
16:28:41  <CIA-53>libuv: build: don't set CC and AR
16:28:41  <CIA-53>libuv: Allows user to override compiler and archiver through the CC and AR environment
16:28:41  <CIA-53>libuv: variables. Note that this already worked as expected with `make CC=clang`. - http://git.io/uhpEkA
16:29:53  <isaacs>bnoordhuis: yeah, that'll probably cause it to blow an assert and crash
16:30:39  <isaacs>bnoordhuis: we could probably ref those thing explicitly
16:30:46  <isaacs>bnoordhuis: but, otoh... don't do that :)
16:30:51  <bnoordhuis>isaacs: yes, we should
16:30:58  <bnoordhuis>but i don't think it'll assert
16:31:26  <isaacs>bnoordhuis: can you get it to do that?
16:31:27  <bnoordhuis>the thread pool worker doesn't know the buffer has gone away so it'll happily write over the memory, never mind that it's no longer owned by that buffer
16:31:29  <bnoordhuis>sure
16:31:59  <isaacs>bnoordhuis: but since we allocate buffers in slabs anyway, wont' it still not get reused for a while
16:32:00  <isaacs>?
16:32:14  * ericktjoined
16:32:49  <isaacs>i mean, it might segfault if that block of memory is no longer node's
16:32:51  <bnoordhuis>err... that's not a safe assumption to make
16:32:56  <isaacs>nono, of course not
16:33:28  <isaacs>another option would be to put the buffer obj and the callback on the reqwrap object explicitly
16:33:41  <bnoordhuis>yep, make it Persistent
16:33:43  <bnoordhuis>or Ref() it
16:33:54  <isaacs>i believe ryah said that was not ideal for some reason. don't recall why.
16:34:35  <bnoordhuis>it has a negative impact on gc performance
16:34:55  * piscisaureusquit (Ping timeout: 248 seconds)
16:36:14  * sh1mmerjoined
16:38:03  * casper__quit (Quit: Leaving)
16:43:54  * piscisaureusjoined
16:48:09  <ryah>?
16:49:25  <ryah>so - should i reland V8 3.6.6?
16:49:30  <ryah>or should we ait for 3.6.7?
16:51:09  <ryah>http://blog.nodejs.org/2011/10/10/node-v0-5-9/#comment-1182 <-- what does this mean?
16:52:21  <ryah>^-- piscisaureus, igorzi_ ?
16:52:31  * piscisaureusquit (Ping timeout: 248 seconds)
16:59:33  <igorzi_>ryah: i have no idea.. he wants to access serial port from node?
17:00:22  <bnoordhuis>ryah: let's wait? unless you feel like tracking down spurious crashes
17:02:15  <ryah>bnoordhuis: are you against removing cmake for some reason?
17:02:35  <bnoordhuis>ryah: not if webos really is dead
17:02:44  <ryah>they can maintain it themselves
17:02:50  <ryah>and or port to GYP
17:03:33  <CIA-53>node: Ryan Dahl * r6a37183 / (14 files in 2 dirs):
17:03:33  <CIA-53>node: Remove cmake build system
17:03:33  <CIA-53>node: This removes one of the three build systems. CMake has never been officially
17:03:33  <CIA-53>node: supported and is not updated with WAF and GYP. WAF itself is considered
17:03:33  <CIA-53>node: legacy and will be removed once the GYP system is feature complete on
17:03:33  <CIA-53>node: Unixes. GYP is used currently on Windows/MSVS. - http://git.io/G8czqg
17:04:44  <ryah>https://github.com/joyent/node/issues/1857 <-- mingw build error
17:05:38  <igorzi_>ryah: i'll get that fixed
17:06:10  <igorzi_>can you please pseudo-assign the issue to me?
17:07:58  <CIA-53>node: Ryan Dahl * ra29421b / CMakeLists.txt : Delete another CMake file - http://git.io/Q0k1vg
17:08:27  <CIA-53>node: Ryan Dahl * r4f4f6df / TODO : Remove old TODO list - http://git.io/K03O9Q
17:09:08  <CIA-53>node: Ryan Dahl * rb280a2d / TODO.win32 : Remove ANOTHER old TODO list - http://git.io/uVp8vg
17:28:32  * bradleymeckjoined
17:29:26  <CIA-53>node: Tomasz Janczuk * r6fadbec / (7 files in 2 dirs):
17:29:26  <CIA-53>node: wix project to create msi installer for windows
17:29:26  <CIA-53>node: Fixes #1706. - http://git.io/Dmb8Lg
17:30:58  * AvianFluquit (Quit: Leaving)
17:31:37  <bnoordhuis>ryah: call?
17:33:21  <ryah>bnoordhuis: yeah
17:34:15  <pquerna>effective October 12th, 2011, Rackspace will stop the sale of Windows Server 2003 images on Cloud Servers
17:34:19  <pquerna>vixtory <3
17:34:26  <pquerna>*victory
17:45:37  <CIA-53>libuv: Igor Zinkovsky * r4ac1309 / (include/uv-private/uv-win.h src/win/winsock.h): fix mingw build - http://git.io/UYH_Mg
17:45:44  <igorzi_>ryah: ---^
17:48:34  <indutny>pquerna: congrats
17:48:35  <ryah>igorzi_: cool thanks
17:50:53  <bnoordhuis>pquerna: you're switching back to windows me?
17:51:30  <pquerna>most were already on 2k8 for cloud anyways, but just eliminates one more buildbot i need
17:52:21  <CIA-53>node: Ryan Dahl * r728b040 / (3 files in 3 dirs):
17:52:21  <CIA-53>node: Upgrade libuv to 4ac1309
17:52:21  <CIA-53>node: Fixes #1857 - http://git.io/DVBk5A
18:03:36  <bnoordhuis>igorzi_ ryah: don't we have the ms call now?
18:04:39  <igorzi_>bnoordhuis: yep
18:05:21  * bnoordhuisdials in
18:16:56  * isaacsquit (Quit: isaacs)
18:21:13  * isaacsjoined
18:28:15  <CIA-53>node: Ryan Dahl * r53c163c / (3 files in 3 dirs): Disable test-child-process-custom-fds.js - http://git.io/7Ivk0g
18:35:45  * sh1mmerquit (Quit: sh1mmer)
18:39:55  * sh1mmerjoined
18:42:55  * AvianFlujoined
18:47:20  <CIA-53>libuv: Igor Zinkovsky * r40b64a8 / src/win/fs.c : windows: don't alloc readdir buffer for empty dirs - http://git.io/J2wsbg
18:47:41  * ryahtopic: https://github.com/joyent/node/wiki/API-changes-between-v0.4-and-v0.6 please help fill this in
18:55:41  <indutny>yeah!
18:55:45  <indutny>0.6
18:57:30  <igorzi_>ryah: we need to add documentation for fs.watch, right?
18:58:55  <ryah>igorzi_: oh gosh - i thought we had that!
18:59:01  <ryah>igorzi_: yes
19:02:29  <bnoordhuis>igorzi_: can you review this change to uv_process_kill? https://github.com/bnoordhuis/libuv/compare/kill-zero
19:05:35  <indutny>ryah: debugger-repl test works fine for me
19:05:39  <indutny>again :D
19:06:16  <CIA-53>node: Ben Noordhuis * r49c3a91 / doc/api/child_processes.markdown : docs: fix typo in code snippet - http://git.io/_rE7nA
19:10:28  <igorzi_>bnoordhuis: lgtm.. can you pls preserve that comment about killed processes normally returning 1?
19:11:08  <bnoordhuis>igorzi_: sure, thanks
19:20:04  <indutny>ryah:
19:20:05  <indutny>https://github.com/joyent/node/pull/1859
19:20:12  <indutny>docs for async readline and repl
19:20:21  <indutny>and also fixed native modules doc
19:26:45  <CIA-53>libuv: Roman Shtylman * ra3d1f6f / (6 files in 5 dirs):
19:26:45  <CIA-53>libuv: add uv_udp_set_membership for unix multicast support
19:26:45  <CIA-53>libuv: - test-udp-multicast-join tests that multicast packets can be received
19:26:45  <CIA-53>libuv: - stub src/win/udp.c until support added - http://git.io/JoQUpw
19:37:57  <CIA-53>libuv: Ben Noordhuis * racc98ca / (test/test-list.h test/test-spawn.c): test: uv_process_kill(proc, 0) should not kill the process - http://git.io/AbFPHw
19:37:57  <CIA-53>libuv: Ben Noordhuis * ra3d495c / src/win/process.c : win: make uv_process_kill(proc, 0) report process status - http://git.io/FhXeDQ
19:39:13  * creationix|workjoined
19:42:15  <bradleymeck>ryah are you trying for C++ addons on windows for 0.6.0? I have a working copy, just I dont know gyp for how to setup the tooling (windows does have an issue where templated classes are wonky though until the linker situation is figured out). and is dlopen being moved into libuv?
19:45:08  <CIA-53>node: Fedor Indutny * r178e2ce / (3 files): docs: improve addons, readline, repl documentation - http://git.io/S7O65A
19:45:13  <indutny>great
19:50:54  <ryah>bradleymeck: i want to give people instructions for how to write their build scripts with gyp - and maybe ship gyp with node - but we're not going to try to get everyone to port their modules
19:51:09  <ryah>we'll do that for v0.8
19:51:14  <bradleymeck>ok
19:51:52  <ryah>it might be nice to have a "node build" command which calls out to GYP
19:52:27  <ryah>bradleymeck: you had to make some patches to export some symbols?
19:53:42  <bradleymeck>a little, the big thing though is V8EXPORT patch to v8.h ... its a one liner but we need to __declspec(dllexport) when static linking into node.exe for v8
19:54:42  <bradleymeck>and I added a NODE_EXPORT thing + the windows.h code
20:02:12  <ryah>bnoordhuis: we'll remove legacy backends after v0.5.10 release
20:07:40  <ryah>or should we just do it now?
20:07:40  <bnoordhuis>why wait?
20:07:40  <ryah>yeah - i guess - it's not like we're going to go back :)
20:07:40  <ryah>and painful to maintain
20:07:41  <CIA-53>node: Ryan Dahl * rd2b8037 / (test/disabled/test-sendfd.js test/simple/test-sendfd.js): disable test-sendfd - http://git.io/KoEV_w
20:12:18  <CIA-53>node: Ryan Dahl * rbe0bb2d / (15 files in 3 dirs): Remove net_legacy timers_legacy - http://git.io/zu1IWQ
20:12:41  <creationix|work>yay, cleaning
20:16:54  <CIA-53>node: Ryan Dahl * r96e423a / (10 files in 3 dirs): Remove child_process_legacy - http://git.io/IMttGQ
20:19:56  <ryah>indutny:
20:19:57  <ryah>=== release test-debugger-repl ===
20:19:57  <ryah>Path: simple/test-debugger-repl
20:19:57  <ryah>Error: timeout!
20:19:57  <ryah>Command: out/Release/node /Users/ryan/projects/node/test/simple/test-debugger-repl.js
20:20:00  <ryah>:<
20:21:57  <CIA-53>node: Ryan Dahl * r58e892d / (10 files in 3 dirs): Remove dns_legacy - http://git.io/g97BBg
20:26:59  <CIA-53>node: Ryan Dahl * r71cce3f / (5 files in 3 dirs): Remove dgram_legacy - http://git.io/OT5JOw
20:28:50  <indutny>ryah: oh no
20:28:55  <indutny>not again
20:28:55  <indutny>:D
20:28:59  <indutny>one second :)
20:29:16  <indutny>ryah: is it persistent?
20:29:35  <indutny>some errors my occur because of v8
20:29:45  <indutny>so probably we should exclude this test from make test
20:29:53  <indutny>and use it just before releases
20:30:36  <indutny>ok, it's not working for me too
20:30:45  <indutny>and then working
20:31:01  <indutny>oh..
20:31:33  <indutny>ryah: I seen that `node debug` was not connecting correctly to v8's debug agent some time ago
20:31:41  <indutny>ryah: I wasn't able to find any reproducing test for it
20:32:28  <indutny>and it looked much more like some v8 bug, because from node's side everything looks similar to normal debug connection (i.e. no errors, just v8 wasn't sending any data
20:32:33  <indutny>)
20:43:15  <CIA-53>node: Ryan Dahl * rcdf5d91 / (12 files in 3 dirs): Remove tty_legacy - http://git.io/uHJcUw
20:46:00  <indutny>ryah: have you thought about using mksnapshot
20:46:01  <indutny>?
20:48:46  <indutny>I think we may significally reduce startup time if we'll amend snapshot with or lib/*.js files
20:49:05  <indutny>and even more they'll be automatically optimized
20:54:14  <ryah>indutny: we use it
20:54:25  <indutny>mksnapshot or js2c?
20:54:55  <indutny>I see js2c in wscript
20:55:04  <indutny>and no trace of mksnapshot
20:55:36  <ryah>we can't snapshot our libraries
20:55:39  <ryah>i think
20:55:49  <indutny>why not
20:55:56  <indutny>v8 is doing so for it's *.js files
20:56:01  <indutny>passing them to js2c first
20:56:06  <indutny>and then creating snapshot
20:56:21  <indutny>I think we can incept into that process
20:56:21  <bnoordhuis>v8's js files don't depend on c++ bindings, i think
20:56:30  <indutny>bnoordhuis: good point
20:56:34  <indutny>hm...
20:56:51  <indutny>yes, really good point
20:56:54  <indutny>:)
20:57:05  <indutny>but we can use %BindingFunction
20:57:06  <ryah>yes, bindings
20:57:08  <indutny>like v8 do
20:57:41  <indutny>I'm not saying that it's essential
20:57:50  <indutny>but just interested if someone ever tried to do that before
20:58:16  <indutny>will it give us performance boost?
21:00:48  * mikealjoined
21:01:23  <bnoordhuis>don't think so - modules are loaded on demand now, a snapshot is all-or-nothing
21:01:37  <indutny>right
21:01:43  <indutny>but they're highly optimized AFAIK
21:01:58  <ryah>runtime functions in V8 - it might be possible somehow
21:02:09  <ryah>but it's probably not easy
21:02:12  <ryah>we'd need to hack v8
21:02:22  <ryah>this is not part of thier API
21:02:26  <indutny>yes
21:02:33  <indutny>we need to hack into build process, actually
21:02:36  <indutny>mostly
21:02:50  <indutny>and some runtime functions hacks, of course
21:06:08  * mikealquit (Quit: Leaving.)
21:07:25  <indutny>well, I think it's so hard that it's mostly impossible
21:07:25  <indutny>:D
21:07:48  <indutny>had a look at runtime.h and runtime.cc and I don't see a simple way to get inside that process
21:07:59  * mikealjoined
21:08:00  <indutny>everything seems to be hardcoded here
21:10:04  <indutny>ok, gtg
21:10:06  <indutny>going to sleep
21:10:07  <indutny>ttyl
21:10:22  <ryah>indutny: night
21:15:16  <igorzi_>bradleymeck: can i see the changes that you're making for NODE_EXPORT?
21:16:35  * isaacsquit (Quit: isaacs)
21:16:45  * mikealquit (Quit: Leaving.)
21:17:01  <bradleymeck>https://github.com/bmeck/node/blob/windowsdll/deps/v8/include/v8.h#L60 , just made it so that I can use __declspec(dllexport) when not using v8 as a shared lib, this should probably be a build option, but i was just doing prototyping to get it to work
21:18:21  <bradleymeck>igorzi_ ^
21:18:50  <ryah>http://herdingcode.com/?p=347 <-- bert talking about the windows port
21:19:05  <igorzi_>bradleymeck: what about exporting libuv?
21:19:52  <bradleymeck>i didnt check libuv bindings in my module, should be the same process as adding NODE_EXPORT
21:20:49  <CIA-53>libuv: Ben Noordhuis * r61343ec / (include/uv.h src/unix/error.c src/uv-common.c): common: add UV_ENOSYS error code - http://git.io/D_415A
21:20:50  <CIA-53>libuv: Ben Noordhuis * r04daabd / src/win/udp.c : win: unbreak build, use UV_ENOSYS - http://git.io/5PSjhg
21:20:50  <bradleymeck>https://github.com/bmeck/node/blob/windowsdll/src/node.h
21:21:28  <bradleymeck>igorzi_ beware though, you cannot use template classes w/ the windows loader right now
21:22:02  <igorzi_>bradleymeck: getting libuv exported is necessary for addons that need to hook into the event loop
21:22:21  <bradleymeck>yep, its not hard, just busy w/ work work past week
21:26:12  <CIA-53>node: Ryan Dahl * r4f59041 / (node.gyp src/node.cc): Fix windows build - http://git.io/XQ48Ww
21:26:38  <bradleymeck>igorzi_, the hard part is figuring out how to support templated classes and being able to check them against node.exe rather than having people give out dlls that barf because they were compiled against a different node version / compilation
21:26:50  <bradleymeck>the exporting of symbols beyond that is simple
21:35:48  <ryah>igorzi_: are you going to do the docs for fs.watch?
21:36:01  <igorzi_>ryah: yes
21:36:03  <ryah>(or should i?)
21:36:04  <ryah>okay
21:38:07  * mikealjoined
21:38:47  <bradleymeck>igorzi_, let me know if you have a way to check dll versions against node versions without big changes, ill be working on windows this weekend probably
21:38:54  <bradleymeck>gtg back to work
21:49:56  <CIA-53>node: Ryan Dahl * rc8dbaf5 / (src/node.cc src/node.js): Remove --use-legacy and --use-uv - http://git.io/tbcUcA
21:49:56  <CIA-53>node: Ryan Dahl * r9631e29 / (3 files in 2 dirs): Remove tests relying on !process.features.uv - http://git.io/dzpaLA
21:49:56  <CIA-53>node: Ryan Dahl * r0e2bbac / test/simple/test-module-load-list.js : Fix test-module-load-list.js - http://git.io/opgDcA
21:53:30  <ryah>% git diff v0.5.9 --shortstat
21:53:30  <ryah> 76 files changed, 3070 insertions(+), 10674 deletions(-)
21:53:34  <ryah>\o/
21:53:47  <igorzi_>bert is a good interviewee :)
21:54:20  <ryah>yeah!
22:00:14  <CIA-53>node: Ryan Dahl * r83fce75 / (4 files): Remove stdio binding source - somehow this was missed in cdf5d91 - http://git.io/JUQR_g
22:18:00  <CIA-53>node: Ryan Dahl * r7085a04 / src/node.cc : Remove process.ENV - http://git.io/M1tMDg
22:20:33  <CIA-53>node: Ryan Dahl * re0b59e7 / src/node.cc : Remove more dead code - http://git.io/4a-EZA
22:27:42  * mralephquit (Quit: Leaving.)
22:28:49  * isaacsjoined
22:31:54  <bnoordhuis>ryah: http://nodejs.org/docs/latest/api/ <- 0.5.6
22:32:45  <ryah>bnoordhuis: you have the power to fix it :)
22:32:54  <bnoordhuis>cool, will do
22:37:38  <igorzi_>ryah bnoordhuis: which OSes (besides windows) support filename in the fs watcher callback? is it just linux and mac?
22:38:00  <bnoordhuis>igorzi_: only linux
22:41:02  * bnoordhuisfinally got gdb to build on sunos!
22:44:46  * isaacsquit (Read error: Connection reset by peer)
22:44:58  * isaacsjoined
22:45:33  <ryah>bnoordhuis: we're still getting reports of problems on lion - were you able to build there?
22:45:38  <ryah>i still haven't upgrade my box
22:46:07  <bnoordhuis>ryah: yes, builds - some tests failing
22:46:30  <bnoordhuis>but not total "won't run at all" behaviour like some people reported
22:54:57  * isaacsquit (Read error: Connection reset by peer)
22:55:05  * isaacsjoined
22:55:19  <rmustacc>I wonder if people have weird messed up libraries with C++ abi woes.
23:01:25  <ryah>how do i use git-blame to see when a line first appeared?
23:05:23  * bradleymeckquit (Ping timeout: 256 seconds)
23:17:52  <ryah>sigh
23:18:10  <ryah>we need to get process.stderr to flush to on each write...
23:18:56  <bnoordhuis>ryah: you should probably use `git log -Sregex <file>` instead of `git blame`
23:19:52  <ryah>im heading home. i'll bb on in a hour or two.
23:21:41  <bnoordhuis>tty tomorrow then, ryah - i hope to be asleep by then :)
23:23:58  <igorzi_>ryah: https://github.com/igorzi/node/commit/6ae1fd4a16fc31d408f7b2e9059dafc1a5ce29cd
23:24:49  * isaacsquit (Read error: Connection reset by peer)
23:50:38  <CIA-53>libuv: Ben Noordhuis * r7b01ad1 / test/test-get-memory.c :
23:50:38  <CIA-53>libuv: test: don't assert that total_mem > free_mem
23:50:38  <CIA-53>libuv: Assertion fails when running in a resource container, like a Solaris zone.
23:50:38  <CIA-53>libuv: The total reported is the container's hard limit, the free memory that of
23:50:38  <CIA-53>libuv: the whole system. - http://git.io/pPJ0CQ
23:55:45  <rmustacc>It seems like ^ should be a bug and that the way we're getting the free memory is wrong.
23:56:25  <rmustacc>Actually, the more I look at that code, the more busted it appears (the sun version of getting memory)
23:59:49  <bnoordhuis>rmustacc: i'm open to suggestions