Index

Show enters and exits. Hide enters and exits.

01:33:20boyscoutAdd Rubinius::Agent and agent loopback connection - b9d29f0 - Evan Phoenix
01:45:28boyscoutCI: rubinius: b9d29f0 successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
07:18:26boyscoutAllow loopback agent to bind to a TCP port - 0051012 - Evan Phoenix
07:20:44boyscoutCI: Commit 0051012 failed. http://github.com/evanphx/rubinius/commit/0051012210bb00eeac948c9f45c2ba553bf54acc
07:25:02dbussinkevan: yay, rails generation bug fixed :)
07:25:09dbussinkevan: you earned your pack of stroopwafels ;)
15:17:34boyscoutAdd missing include. - a4d4c20 - Brian Ford
15:33:06boyscoutCI: rubinius: a4d4c20 successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
17:01:24dbussinkevan: saw you snuck in a talk at railsconf :P
17:01:33evan:)
17:09:56dbussinkevan: btw, been testing a bit with rails3, but i see exceptions fly by sometimes in the stdout log
17:10:03dbussinklet me get one
17:10:05evank
17:11:12dbussinkevan: https://gist.github.com/fc28c4ba796e442b2ec6
17:11:47evanhrm. ok.
17:13:29evandbussink: well, any investigation you can do would be great :)
17:13:50evanactually... hm.
17:14:20evandoes webrick use nonblocking IO?
17:19:15dbussinkevan: dunno, i can take a look at the code
17:21:09evank
17:26:06dbussinkevan: it does try to do non blocking stuff
17:26:43dbussinkevan: http://github.com/evanphx/rubinius/blob/master/lib/webrick/utils.rb#L21
17:28:15evanok, so
17:28:27evanshould a IO marked as nonblock not be raising EAGAIN in write?
17:32:27dbussinkevan: dunno actually
17:47:02evanman
17:47:11evanmaybe I should just release 1.0.1 from master.
17:47:28evani'm having to cherry pick 90+% of the commits
17:47:30dbussinkevan: i've been meaning to ask that question too
17:47:46dbussinkthat might be just as good and a lot less work
17:48:03brixenseems reasonable to me
17:48:05evanthe debugger and agent work are new features
17:48:12evanbut i'll just say they're experimental for 1.0.1
17:48:21brixenyeah
17:48:28evanok, i'm just going to do that.
17:48:34evanwe did so many fixes
17:48:55evancherry picking them is going to create another isuse where people are running a code base that we've never run
17:49:08brixentrue
17:54:03evandbussink: ah, I think I see.
17:54:22evanMRI's IO#write actually handles EAGAIN internally.
17:55:34evanmm, maybe..
17:57:07evanyeah, it does..
17:57:13evanok, we'll need to do that too..
17:57:43dbussinkevan: ah ok, well i don't see anything wrong in the browser actually
17:57:50evanok
17:58:00dbussinkso maybe webrick also handles something
17:58:02evani'm going to go ahead and fix it in IO::write in C++ land.
17:58:07dbussinkok, cool
17:59:14evanah!
17:59:17evanok, i get what MRI does.
17:59:21evanthis is silly
17:59:23evanbut thats fine.
18:01:51dbussinksilly mri!
18:38:03evandbussink: poke
18:38:23dbussinkevan: auch
18:38:30evani'm going to push a fix
18:38:32evancould you test it?
18:38:42dbussinkevan: ok :)
18:38:44evani'm trying to figure out how/if i can write a spec for write() setting EAGAIN
18:40:11boyscoutHandle write() setting EAGAIN properly - 652b7fe - Evan Phoenix
18:48:36boyscoutCI: rubinius: 652b7fe successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
18:51:40dbussinkevan: got a new one :P
18:51:41dbussinkhttps://gist.github.com/cbe903b75897b1da0fa0
18:52:10evanwhy is eof? calling fill_from...
18:52:28evanfuck.
18:52:35evandamn thing.
18:52:53evanwhy the heck is that line in there.
18:53:40dbussinkevan: should go sit together at railsconf, me causing and finding bugs, you fixing them :P
18:53:48evansure!
18:54:55evandbussink: on line 820 of io.rb put
18:55:00evanreturn true if @eof
18:55:10evansee what happens
18:57:53dbussinkevan: replace the line or insert it?
18:58:05evaninsert
18:58:40dbussinkevan: doesn't seem to make a difference
18:58:48dbussinkhave a new exception too now: https://gist.github.com/aca58865ac7529e6b6dd
18:58:48evank
18:59:12evanperhaps I should just repro it locally
18:59:14evanand fix them all.
18:59:41dbussinkevan: let me see how much setup is needed
19:02:43dbussinkevan: repro looks pretty simple, i have it now here with a rails3 beta3 app
19:02:52dbussinkand one scaffolded person model
19:02:52evanok
19:02:58evanput the setup in a gist
19:03:03evanand i'll tackle it after lunch
19:06:54dbussinkevan: http://github.com/evanphx/rubinius/issues/issue/338
19:07:03dbussinkevan: it's even an issue ;)
19:07:05evandanku!
19:07:12evanas it should be!
19:12:43dbussinkevan: are you already at railsconf during the tutorial day?
19:13:02dbussinkthey had like these hacking rooms 2 years ago, dunno if they still have something like that
20:14:03evandbussink: what were you using to connect to webrick?
20:14:08evansafari? chrome? curl?
20:14:16dbussinkevan: chrome and firefox
20:14:20evanhrm
20:14:20evanok
20:14:29evannot getting the exceptions here.
20:14:29dbussinkevan: hitting it harder makes it appear more
20:14:35evank
20:14:39dbussinkso not just wait but cmd r a lot
20:14:40dbussink:)
20:14:52evanoh, there we go.
20:15:04evanlots.times { cmd-r }
20:18:18evanzoinks!
20:19:58dbussinkevan: seeing it?
20:20:03dbussinkevan: or seeing other things? :P
20:20:10evanyep
20:20:10evangot it.
20:20:15evani caused another bug that made the VM crash
20:20:17evanbut i fixed that.
20:20:18evan:)
20:20:32evani'm having the VM treat ECONNRESET the same as read() return 0 bytes (ie, eof)
20:25:23dbussinkevan: you should try hammering it with ab ;)
20:25:31evanok
20:25:38evanmore stuff might pop up.
20:26:28dbussinkevan: not that i tried, but if this crashes stuff, it might expose other things yeah
20:28:00evanman.
20:28:09evanusing ab -c > 1 causes a bunch of rails errors
20:28:16evanbecause of, likely, autoloading.
20:29:41dbussinkugh, nasty
20:29:47dbussinkevan: does mri handle that correctly then?
20:29:53evandunno
20:29:56evanmy guess is no.
20:30:17evanthis is why people shouldn't use autoload.
21:11:37boyscoutHandle ECONNRESET like eof. Fixes #338 - 6fe7cfc - Evan Phoenix
21:19:44boyscoutCI: rubinius: 6fe7cfc successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
22:32:18evantarcieri: awesome about rubinius day-to-day
22:32:18evan:)
22:32:55tarcieriheh
22:33:05tarcieriyeah have rails 3 on it too
22:33:17tarcieriuhoh meeting!
22:33:30evanuhoh!
22:33:31evanrun!!
22:33:39brixenthe other direction!
22:33:57evanoh no, i think tarcieri got captured by the meeting monster.
22:34:05evanbrixen: almost got passenger running on rbx
22:34:19evanwe're missing a couple of socket methods i'm adding
22:34:24brixensweet
22:41:21evanwoo, got it working.
22:43:42evanhttp://github.com/FooBarWidget/passenger/blob/master/lib/phusion_passenger/utils.rb#L37
22:43:48evanthat ^^ line is the only fix
22:51:35brixenawesome
22:53:35tarcieriyay not terribly long meeting
22:55:49evanhe survived the meeting monster!
22:55:57tarcieriyay meeting monster vanquished!
23:18:01toulmeantarcieri: I need some tips on that monster. I tried a combo x + triangle, but...
23:19:51tarcierieach meeting monster is different, and some are more difficult to vanquish than others
23:22:16toulmeanlooks like it. The higher your level is...