#ruby-core on 2013-02-07 — irc logs at freenode.irclog.whitequark.org

00:06 tenderlove has quit [Read error: Connection reset by peer]

00:07 davidbalber|away is now known as davidbalbert

00:14 johndouthat has quit [Quit: johndouthat]

00:26 davidbalbert is now known as davidbalber|away

00:41 idkazuma has quit [Remote host closed the connection]

00:41 idkazuma has joined #ruby-core

00:46 idkazuma has quit [Ping timeout: 264 seconds]

00:59 tenderlove has joined #ruby-core

01:06 <znz_v> biff: [ruby-changes:27071] nobu:r39123 (trunk): test_basicinstructions.rb: fix typo - http://mla.n-z.jp/?ruby-changes=27071

01:07 <znz_v> biff: [ruby-changes:27072] nobu:r39124 (trunk): test_basicinstructions.rb: implementation defined - http://mla.n-z.jp/?ruby-changes=27072

01:19 carlosantonio has quit [Remote host closed the connection]

01:25 _br_ has quit [Excess Flood]

01:28 _br_ has joined #ruby-core

01:29 johndouthat has joined #ruby-core

01:30 _br_ has quit [Excess Flood]

01:32 _br_ has joined #ruby-core

01:40 <znz_v> biff: [ruby-changes:27073] nobu:r39125 (trunk): test_basicinstructions.rb: fix typo - http://mla.n-z.jp/?ruby-changes=27073

01:57 drbrain has quit [Remote host closed the connection]

02:19 nari has joined #ruby-core

02:21 marcandre has quit [Remote host closed the connection]

02:34 drbrain has joined #ruby-core

02:36 drbrain has quit [Remote host closed the connection]

02:36 drbrain has joined #ruby-core

03:28 xibbar has quit [Remote host closed the connection]

04:07 xibbar has joined #ruby-core

04:28 indirect_ has joined #ruby-core

04:29 indirect has quit [Ping timeout: 264 seconds]

04:29 indirect_ is now known as indirect

04:57 nokada has joined #ruby-core

04:57 nokada_ has quit [Read error: Connection reset by peer]

05:01 shirokuro11 has joined #ruby-core

05:30 kosaki8 has joined #ruby-core

06:05 <znz_v> biff: [ruby-changes:27074] drbrain:r39126 (trunk): * lib/rubygems/package.rb: Ensure digests are generated for signing. - http://mla.n-z.jp/?ruby-changes=27074

06:23 oscardelben has joined #ruby-core

06:33 <znz_v> biff: [ruby-changes:27075] nobu:r39127 (trunk): test_libs.rb: new test - http://mla.n-z.jp/?ruby-changes=27075

06:33 <znz_v> biff: [ruby-changes:27076] nobu:r39128 (trunk): mkmf.rb: fix merge_libs - http://mla.n-z.jp/?ruby-changes=27076

06:59 oscardelben has quit [Quit: oscardelben]

07:01 kosaki8 has quit [Ping timeout: 246 seconds]

07:15 charliesome has joined #ruby-core

07:27 vondruch has joined #ruby-core

07:29 <znz_v> biff: [ruby-changes:27077] nobu:r39129 (trunk): mkconfig.rb: fix concatenated lines - http://mla.n-z.jp/?ruby-changes=27077

07:30 <znz_v> biff: [ruby-changes:27078] nobu:r39130 (trunk): Makefile.sub: rubyarchdir in config.status - http://mla.n-z.jp/?ruby-changes=27078

07:40 shirokuro11 has quit [Remote host closed the connection]

08:02 <znz_v> biff: [ruby-changes:27079] nobu:r39131 (trunk): rbinstall.rb: CONFIG['arch'] not $arch - http://mla.n-z.jp/?ruby-changes=27079

08:13 judofyr has joined #ruby-core

08:22 johndouthat has quit [Quit: johndouthat]

08:44 heroux has quit [Ping timeout: 252 seconds]

09:34 <charliesome> _ko1: are you there?

09:34 nari has quit [Ping timeout: 276 seconds]

09:52 <_ko1> charliesome: yes

09:52 <charliesome> howdy!

09:52 <_ko1> hi

09:52 <charliesome> i was investigating ruby performance on a rails app i work on

09:52 <_ko1> cool

09:53 <charliesome> and i noticed that whenever we use OpenStruct, all the method caches are invalidated because it defines methods on its singleton class

09:53 <judofyr> charliesome: yes, I've also noticed *really* bad performance using OpenStruct

09:53 <charliesome> i thought about it for a bit and i think i have an idea to avoid invalidating the cache when defining singleton methods

09:54 <charliesome> we should store an inherited flag on RClass structs that is set to true when it is inherited

09:54 <charliesome> when defining a method, the vm state is only incremented if a) the method definition overrides an existing one or b) the class has subclasses (inherited flag is set)

09:54 <charliesome> _ko1: thoughts?

09:55 <charliesome> judofyr: yeah i was surprised as well

09:55 <_ko1> n-

09:55 <_ko1> not enough

09:56 <charliesome> is there a scenario where this causes incorrect behaviour, or do you not think this case is worthy of optimization?

09:56 <charliesome> because defining singleton methods on individual object is a *very* common operation

09:57 <_ko1> no doubt that it is worth

09:57 <_ko1> wait a morment

09:58 xibbar has quit [Remote host closed the connection]

09:59 <_ko1> it my mistake

09:59 <_ko1> your idea is good

09:59 <_ko1> the last problem is

09:59 <_ko1> set a flag correctly

10:00 <_ko1> set a `inherited' flag

10:01 xibbar has joined #ruby-core

10:03 <_ko1> big 2.1 feature :)

10:03 <_ko1> charliesome: do you have a patch?

10:04 <charliesome> _ko1: i have been experimenting, nothing working yet

10:04 <charliesome> _ko1: would you like me to work on a patch?

10:04 <charliesome> then i can submit it for your review

10:04 <_ko1> if you can :)

10:04 <charliesome> awesome

10:04 _br_ has quit [Excess Flood]

10:04 <_ko1> rails people like OpenStruct?

10:05 <charliesome> _ko1: not sure about rails itself, but openstruct is very commonly used

10:05 <_ko1> wow

10:05 <_ko1> why people don't use Hash

10:05 <_ko1> ...

10:05 <charliesome> _ko1: they like the method call syntax

10:06 <charliesome> i hacked my copy of ruby at work to print a star to the terminal every time the vm state was incremented

10:06 <charliesome> 200 vm state changes per request :\

10:06 <_ko1> LoL

10:06 <charliesome> many were from openstruct

10:06 <_ko1> good consideration

10:07 <_ko1> there are several ideas about inline cache invalidation protocol

10:07 <_ko1> (1) use only vm state (current)

10:07 <_ko1> (2) separate vm state (JRuby use)

10:07 <charliesome> _ko1: i noticed you have future support for only invalidating by a certain ID

10:07 <_ko1> (2) separate vm state by Classes/Modules (JRuby use)

10:07 <charliesome> as in, the function is there but at the moment it just invalidates everything

10:08 <_ko1> (3) separate vm state by method names (Rubinius use)

10:08 <_ko1> (4) more complex one

10:08 <charliesome> unfortunately different invalidation strategies might be better or worse for different apps

10:08 <_ko1> (4) is described at http://dl.acm.org/citation.cfm?id=1869634

10:09 <charliesome> maybe in a future version we can use an env var to control invalidation strategy?

10:09 <charliesome> similar to gc tuning

10:09 <_ko1> charliesome: Yes. absolutely

10:09 _br_ has joined #ruby-core

10:10 <_ko1> what means "an env var to control invalidation "?

10:10 <_ko1> ah

10:10 <_ko1> ENV.

10:10 <_ko1> i understand

10:10 <_ko1> i'm not sure.

10:10 <_ko1> maybe not because invalidation code should be inlined into VM code

10:10 <charliesome> oh yes, that is a valid concern

10:10 <_ko1> should not add additional branch

10:11 <_ko1> build option is acceptable

10:11 <_ko1> for testing.

10:11 <charliesome> also since most people compile their own ruby, people can set the best option for their use

10:12 <_ko1> yes

10:12 <_ko1> it is good theme for under graduate students :)

10:14 <charliesome> _ko1: also, have you considered using -flto on compilers that support it?

10:14 <_ko1> i have no experience

10:14 <_ko1> could you try it?

10:15 <charliesome> i tried it, but there were crashes

10:15 <_ko1> ah.,,

10:15 <_ko1> my understanding

10:15 <charliesome> -flto enables some very aggressive optimizations at link time

10:15 <_ko1> LTO does not change any behavior

10:15 _br_ has quit [Excess Flood]

10:15 <_ko1> i see

10:16 <charliesome> i think ruby relies on a lot of undefined behaviour at the moment

10:16 <_ko1> may be it causes GC problem

10:16 <charliesome> _ko1: i think that might be the case

10:16 <_ko1> inlining some function causes mark miss.

10:17 _br_ has joined #ruby-core

10:21 carlosantonio has joined #ruby-core

10:21 _br_ has quit [Excess Flood]

10:24 _br_ has joined #ruby-core

10:27 xibbar has quit [Remote host closed the connection]

10:45 idkazuma has joined #ruby-core

10:56 carlosantonio has quit [Remote host closed the connection]

11:05 carlosantonio has joined #ruby-core

11:08 <charliesome> why do people use symbols where they should use strings :|

11:23 idkazuma has quit [Remote host closed the connection]

11:24 kosaki8 has joined #ruby-core

11:55 <znz_v> biff: [ruby-changes:27080] usa:r39132 (trunk): * win32/Makefile.sub (config.status): added variables which were - http://mla.n-z.jp/?ruby-changes=27080

11:58 kosaki8 has quit [Ping timeout: 256 seconds]

12:01 travis-ci has joined #ruby-core

12:01 <travis-ci> [travis-ci] The build has errored. by @usa: See http://travis-ci.org/ruby/ruby/builds/4639281

12:01 travis-ci has left #ruby-core [#ruby-core]

12:11 <znz_v> biff: [ruby-changes:27081] usa:r39133 (trunk): * test/rubygems/test_gem_config_file.rb - http://mla.n-z.jp/?ruby-changes=27081

12:22 <znz_v> biff: [ruby-changes:27082] nobu:r39134 (trunk): vpath.rb: no rbconfig.rb - http://mla.n-z.jp/?ruby-changes=27082

12:24 travis-ci has joined #ruby-core

12:24 travis-ci has left #ruby-core [#ruby-core]

12:24 <travis-ci> [travis-ci] The build passed. by @usa: See http://travis-ci.org/ruby/ruby/builds/4639632

12:30 idkazuma has joined #ruby-core

13:07 <znz_v> biff: [ruby-changes:27083] akr:r39135 (trunk): * configure.in: define linker for shared library on MirOS BSD. - http://mla.n-z.jp/?ruby-changes=27083

13:21 vondruch has quit [Quit: Ex-Chat]

13:25 <_ko1> charliesome: because it is more shoter ("foo" -> :foo) :)

13:25 <charliesome> _ko1: perhaps

13:26 <charliesome> in my language, i have symbols internally but not exposed to userland

13:26 <charliesome> i also have a 'string syntax that lets you quote identifiers similar to symbols in ruby

13:31 <_ko1> your language?

13:32 heroux has joined #ruby-core

13:32 <mame1> charliesome: hello

13:33 <charliesome> mame1: howdy

13:33 <charliesome> k

13:33 <mame1> r30148

13:33 <mame1> oops, there is no useful bot

13:33 <charliesome> the env one?

13:34 <mame1> some people complains to your r39004

13:34 <mame1> because it might be an intentional change by matz at r30148

13:34 <mame1> http://svn.ruby-lang.org/cgi-bin/viewvc.cgi?revision=30148&view=revision

13:35 <charliesome> oh

13:35 <mame1> * array.c (rb_ary_dup): should copy contents only. no instance

13:35 <mame1> variable, no class would be copied.

13:36 <nurse> The point is that don't change behavior without matz's accept

13:36 <mame1> however, matz is now saying the opposite opinion against himself

13:36 <charliesome> sorry for changing behaviour

13:36 <mame1> he now prefers the old 1.9.3 behavior

13:36 <charliesome> the bug looked like a regression, but i fixed the wrong place

13:37 <mame1> oops, 1.9.3 behavior -> 1.9.2 behavior

13:38 <mame1> indeed rb_ary_dup impacts too extensively a little

13:39 <mame1> then how do you fix?

13:39 <charliesome> so this was the bug http://bugs.ruby-lang.org/issues/7768

13:40 <mame1> fix only Array#uniq ?

13:40 <charliesome> mame1: that would be the least intrusive way

13:41 <charliesome> mame1: do you want me to do it?

13:42 <mame1> could you create a patch for the way?

13:42 <charliesome> sure, one moment

13:43 <mame1> just confirm: you didn't intend to change the behavior of other methods than Array#uniq, right?

13:43 <mame1> then, it looks good to me to choose "the least intrusive way"

13:44 <charliesome> mame1: i thought that perhaps there were other methods that would behave inconsistently in some cases like #uniq

13:44 <charliesome> but i will give you a patch that undoes it

13:45 <mame1> honestly i agree with you to some extent

13:45 <mame1> but the timing is bad a little

13:45 <charliesome> yes, i agree

13:45 <charliesome> sorry once again

13:46 <charliesome> something to discuss after 2.0 is released maybe?

13:46 <mame1> no prob, thank you for your cooperation

13:46 <mame1> no plan :-)

13:47 <mame1> ah, it may be difficult to change 2.0.0 after 2.0.0-p0 is released

13:47 <charliesome> for next minor, i mean

13:47 <mame1> 2.0.1 or 2.1.0 will include the change, which matz will decide

13:47 <mame1> yes

13:48 <charliesome> mame1: https://gist.github.com/charliesome/adeff49e900f6b8a75fc

13:48 vondruch has joined #ruby-core

13:48 <charliesome> the test i added in r39004 still passes

13:49 <charliesome> so i think this is ok if you're happy to change Array#uniq's behaviour

13:49 <judofyr> why not changing rb_ary_dup?

13:51 <mame1> charliesome: it is really a regression, isn't it?

13:52 <charliesome> mame1: i think so - see http://bugs.ruby-lang.org/projects/ruby-trunk/repository/revisions/26987/diff/array.c

13:52 <charliesome> it looks like an attempt at an optimization

13:52 <charliesome> judofyr: too late to change Array#dup's behaviour

13:52 <judofyr> ah, right

13:52 <mame1> yeah, i've seen the diff, but i've not checked the other region of code

13:53 <charliesome> mame1: maybe the safest option is to revert r39004 completely?

13:53 <mame1> it returns a new object whose class is the same of the argument, say, when its length is >= 2?

13:53 <charliesome> mame1: yes

13:54 <mame1> hmm, thanks. let me consider for a minute :-)

13:54 <charliesome> no problem

13:56 <_ko1> charliesome, mame: could you file this discussion (summary) to bug ticket? we will lost progress only on IRC

13:56 <_ko1> after you finish discussion

13:58 carlosantonio has quit [Remote host closed the connection]

13:59 <charliesome> _ko1: should i open feature ticket for next minor for the singleton method cache optimization?

14:00 <_ko1> it is okay after your patch :)

14:00 <charliesome> ok, so i'll wait until i have a patch ready

14:01 __BigO__ has joined #ruby-core

14:15 <mame1> charliesome: sorry for let you wait

14:15 <mame1> i decided that i'll release rc2 with the *current* behavior

14:15 <charliesome> mame1: current as in trunk or current 1.9.3?

14:16 <mame1> because matz again prefered not only Array#uniq but also other Array methods to return the original class rather than Array

14:16 <mame1> current trunk

14:16 <mame1> iow your r39004 is accepted :-)

14:16 <charliesome> oh cool!

14:17 <mame1> sorry for let you write an extra unnecessary patch

14:17 <charliesome> that's ok, sorry for causing this whole issue in the first place

14:17 <charliesome> next time i'll be more careful

14:18 <mame1> don't mind, it was a good patch as a result :-)

14:19 <mame1> thanks again

14:19 <charliesome> i'll post the summary to #7768 as per _ko1's request

14:21 <_ko1> mame1: no need to post summary?

14:22 <charliesome> _ko1: oh well, posted summary anyway :p

14:22 <charliesome> it may be useful, it may not be

14:22 <mame1> it look a good idea to leave an evidence

14:26 <charliesome> mame1: btw i am excited for 2.0.0's release

14:28 <mame1> maybe it will be tomorrow unless there is no other incidents...

14:29 <_ko1> charliesome: :) thanks

14:30 <charliesome> anyway, good night

14:30 <_ko1> mame1: no plan delay?

14:30 charliesome has quit [Quit: Textual IRC Client: www.textualapp.com]

14:31 <mame1> _ko1: it has been already delayed enough

14:32 <_ko1> not an rc2, but about 2.0.0p0

14:32 <mame1> no plan

14:32 <mame1> do you think it should be delayed?

14:32 <_ko1> i don't have any idea

14:34 <mame1> i didn't know that charliesome is in as almost same time zone as Japan

14:38 <_ko1> me too

15:01 davidbalber|away is now known as davidbalbert

15:04 carlosantonio has joined #ruby-core

15:09 <znz_v> biff: [ruby-changes:27084] svn:r39137 (trunk): * 2013-02-08 - http://mla.n-z.jp/?ruby-changes=27084

15:11 <znz_v> biff: [ruby-changes:27085] akr:r39136 (trunk): * process.c (obj2gid): use getgrnum_r() only if getgrnam_r() and - http://mla.n-z.jp/?ruby-changes=27085

15:20 judofyr has quit [Ping timeout: 264 seconds]

15:28 judofyr has joined #ruby-core

15:45 oscardelben has joined #ruby-core

15:58 oscardelben has quit [Quit: oscardelben]

16:03 <znz_v> biff: [ruby-changes:27086] nagachika:r39138 (trunk): * ChangeLog: fix a typo for r39136. - http://mla.n-z.jp/?ruby-changes=27086

16:16 vondruch has quit [Quit: Ex-Chat]

16:16 oscardelben has joined #ruby-core

16:29 judofyr has quit [Remote host closed the connection]

16:39 johndouthat has joined #ruby-core

16:49 oscardelben has quit [Quit: oscardelben]

17:11 idkazuma has quit [Remote host closed the connection]

17:32 carlosan_ has joined #ruby-core

17:34 carlosantonio has quit [Ping timeout: 248 seconds]

17:59 _br_ has quit [Excess Flood]

18:00 _br_ has joined #ruby-core

18:01 judofyr has joined #ruby-core

18:04 _br_ has quit [Excess Flood]

18:06 _br_ has joined #ruby-core

18:15 marcandre has joined #ruby-core

18:31 __BigO__ has quit [Remote host closed the connection]

18:40 __BigO__ has joined #ruby-core

18:47 carlosantonio has joined #ruby-core

18:49 carlosan_ has quit [Ping timeout: 256 seconds]

19:16 wakoinc has joined #ruby-core

19:16 wakoinc has quit [Changing host]

19:16 wakoinc has joined #ruby-core

19:23 drbrain has quit [Remote host closed the connection]

19:46 _br_ has quit [Excess Flood]

19:49 wakoinc has quit [Read error: Connection reset by peer]

19:49 wakoinc has joined #ruby-core

19:50 judofyr has quit [Remote host closed the connection]

19:51 _br_ has joined #ruby-core

19:52 _br_ has quit [Excess Flood]

19:57 _br_ has joined #ruby-core

20:15 tenderlove has quit [Remote host closed the connection]

21:09 johndouthat has quit [Quit: johndouthat]

21:12 johndouthat has joined #ruby-core

21:12 johndouthat has quit [Remote host closed the connection]

21:22 drbrain has joined #ruby-core

21:25 tenderlove has joined #ruby-core

21:52 kosaki8 has joined #ruby-core

22:00 __BigO__ has quit [Remote host closed the connection]

22:04 wakoinc has quit [Quit: wakoinc]

22:08 tenderlove has quit [Read error: Connection reset by peer]

22:09 tenderlove has joined #ruby-core

22:09 hone has quit [Ping timeout: 276 seconds]

22:15 hone has joined #ruby-core

22:27 hone has quit [Ping timeout: 255 seconds]

22:27 hone has joined #ruby-core

22:32 idkazuma has joined #ruby-core

22:32 hone has quit [Ping timeout: 245 seconds]

22:39 hone has joined #ruby-core

22:49 __BigO__ has joined #ruby-core

22:54 xibbar has joined #ruby-core

22:54 drbrain has quit [Remote host closed the connection]

22:55 <znz_v> biff: [ruby-changes:27087] drbrain:r39139 (trunk): * lib/rubygems/dependency_installer.rb: Only install local gems if - http://mla.n-z.jp/?ruby-changes=27087

23:01 <znz_v> biff: [ruby-changes:27088] zzak:r39141 (trunk): * lib/net/http.rb (HTTP.post_form): Fix module scope in documentation - http://mla.n-z.jp/?ruby-changes=27088

23:02 <znz_v> biff: [ruby-changes:27089] zzak:r39140 (trunk): * compar.c (cmp_equal): Document ignored exception and return false - http://mla.n-z.jp/?ruby-changes=27089

23:21 <znz_v> biff: [ruby-changes:27090] drbrain:r39142 (trunk): * lib/rubygems/package/old.rb: Disallow installation of old-format - http://mla.n-z.jp/?ruby-changes=27090

23:23 nari has joined #ruby-core

23:37 drbrain has joined #ruby-core

23:42 __BigO__ has quit [Remote host closed the connection]

23:57 nari has quit [Ping timeout: 276 seconds]