Index

Show enters and exits. Hide enters and exits.

07:18:27elliotcmMorning all.
07:40:23parndtmorning elliotcm
17:31:58boyscoutGet jit.inline.blocks running again - f174b48 - Evan Phoenix
17:31:58boyscoutUse raw Tuple access in Array#each - dbe0373 - Evan Phoenix
17:40:55boyscoutCI: rubinius: dbe0373 successful: 3441 files, 13524 examples, 41038 expectations, 0 failures, 0 errors
17:57:38dbussinkevan: doing some performance things?
17:57:48evanI was this weekend, yeah.
17:58:04evanjit.inline.blocks made a Array#each benchmark go from 6s to 2s :)
18:01:25dbussinkevan: hehe, cool :)
18:03:05kronos_vanoevan, Awesome!
18:03:20dbussinkevan: does that different each code help a lot too? or does it make it easier on the inlining?
18:03:32evanit helps, yeah.
18:03:34evanit's simpler
18:03:43evanso the JIT can see more operations.
18:03:52dbussinki can imagine that yeah
18:04:05evani don't want to replace all the to_iter with raw @tuple access though
18:04:09evanbut for #each, it makes sense.
18:04:44dbussinkevan: hmm, does this also means that now as a side effect the behavior is that added elements to the array are iterated over?
18:05:01dbussinkevan: which is also mri behavior (but not specificied behavior though)
18:05:25evanthat's true for to_iter as well.
18:05:34evanthis code has the same behavior as to_iter
18:05:59evanbecause the Array::iterator holds a ref to the @tuple, @total, @start, and the current index
18:06:31ezmobiusdoes rake install work reliably now or should i stay away from it still?
18:06:32evanadding elements to an array during iteration and seeing them is undefined.
18:06:36dbussinkevan: but isn't that ref copied for the iterator?
18:06:39dbussinkevan: the value i mean
18:06:42evanezmobius: it should be working ok.
18:06:47ezmobiuscool
18:06:48evandbussink: which?
18:06:59dbussinkevan: let me check something
18:08:14dbussinkevan: well, now .each uses @total directly, if it's passed to an iterator, changing @total doesn't change the value passed to the iterator when it was constructed
18:08:26evanshow me
18:08:28evani don't follow.
18:08:42evanwhat is "the value" ?
18:10:32dbussinkevan: hmm, nvm, i was rambling something that isn't right :0
18:10:33dbussink:)
18:10:59evani'm glad I asked you to verify before explaining then
18:11:00evan:)
18:11:19dbussinkevan: well, i was writing up a code sample to see if it was actually doing what i thought it was
18:11:21dbussinkbut it isn't
18:11:36evanah.
18:12:11dbussinkevan: ah, wait, i see, i was looking at it wrong, @total is not evaluated again in each iteration
18:12:20evanno
18:12:22evanit's not.
18:14:35dbussinkevan: btw, did you get to investigating that rails app generation issue?
18:14:46evani am going to today
18:18:13dbussinkevan: ah, cool :)
18:18:27dbussinkevan: i'll add another stroopwafel pack to my bag for that ;)
18:18:40evanWOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
18:18:43evan:)
19:31:30boyscoutTag spec failing due to IPv6 effects on linux - 3ddf2c1 - Evan Phoenix
19:32:59boyscoutLet --enable-llvm do something. Fixes #336. - a5f79ac - Evan Phoenix
19:39:32boyscoutCI: rubinius: 3ddf2c1 successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
21:19:09boyscoutAdd flag to disallow direct allocation. Fixes #335. - 72cc050 - Evan Phoenix
21:31:06boyscoutCI: rubinius: 72cc050 successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
21:44:16evanoh goody
21:44:19evanmore Dir.glob bugs.
21:44:20evan:/
23:25:44boyscoutAdd flag support to new Dir.glob, always use it - 69c2000 - Evan Phoenix
23:25:44boyscoutRemove the old Dir#glob entirely - 78c43ac - Evan Phoenix
23:41:05evanhehe
23:41:25evanruby-core added a .gdbinit
23:41:31evanwith the same macro names as we use
23:41:37evani guess they've been checkin' us out!
23:48:33BrianRice-workoh, maybe slate should crib from that, too
23:49:14evan:)
23:49:24evanthey didn't actually reuse anything
23:49:30evanjust called their main macro 'rp'
23:49:38boyscoutCI: rubinius: 78c43ac successful: 3441 files, 13523 examples, 41036 expectations, 0 failures, 0 errors
23:49:39BrianRice-workoh I understand. we're just not aliasing anything yet
23:49:41evanheck, it's so obvious, they might have never seen the rubinius one.
23:50:05BrianRice-work(I don't think we've needed gdb in months, thankfully)
23:50:15evanlucky you!