[#50466] [ruby-trunk - Bug #7492][Open] Segmentation fault at DL::TestDL#test_call_double on x64 Windows 8 — "phasis68 (Heesob Park)" <phasis@...>
23 messages
2012/12/02
[#59083] [ruby-trunk - Bug #7492] Segmentation fault at DL::TestDL#test_call_double on x64 Windows 8
— "phasis68 (Heesob Park)" <phasis@...>
2013/12/13
[#50483] [IMPORTANT] 2.0.0 release plan — Yusuke Endoh <mame@...>
ALL COMMITTERS SHOULD READ THIS MAIL! コミッタはこのメール読んで！
5 messages
2012/12/02
[#50561] [ruby-trunk - Bug #7513][Open] TracePoint#enable/disable should not cause error — "ko1 (Koichi Sasada)" <redmine@...>
7 messages
2012/12/05
[#50575] [ruby-trunk - Feature #7517][Open] Fixnum::MIN,MAX — "matz (Yukihiro Matsumoto)" <matz@...>
20 messages
2012/12/05
[#50593] [ruby-trunk - Feature #7517] Fixnum::MIN,MAX
— "shyouhei (Shyouhei Urabe)" <shyouhei@...>
2012/12/05
[#50594] Re: [ruby-trunk - Feature #7517] Fixnum::MIN,MAX
— Charles Oliver Nutter <headius@...>
2012/12/05
On Wed, Dec 5, 2012 at 12:24 PM, shyouhei (Shyouhei Urabe)
[#50636] [ruby-trunk - Bug #7528][Open] CSV.== fails to check object type — "SteveW (Stephen Wattam)" <stephenwattam@...>
6 messages
2012/12/06
[#50660] [ruby-trunk - Feature #4085] Refinements and nested methods — "trans (Thomas Sawyer)" <transfire@...>
3 messages
2012/12/07
[#50699] Commit access for Yehuda Katz — Aaron Patterson <tenderlove@...>
Hi,
4 messages
2012/12/08
[#50923] Re: Commit access for Yehuda Katz
— Charles Oliver Nutter <headius@...>
2012/12/16
I will +1 this, unsure if it has happened already (it's "catch up on
[#50733] [ruby-trunk - Bug #7539][Open] Misleading error message "can't convert nil into string" — "connec (Chris Connelly)" <chris@...>
8 messages
2012/12/10
[#50755] Becoming a committer — Charlie Somerville <charlie@...>
Hi ruby-core,
21 messages
2012/12/11
[#50759] Re: Becoming a committer
— Yukihiro Matsumoto <matz@...>
2012/12/11
Hi,
[#50784] Re: Becoming a committer
— Charles Oliver Nutter <headius@...>
2012/12/11
It's really this easy? If so, I'll send over my public key today :)
[#50795] Re: Becoming a committer
— Yukihiro Matsumoto <matz@...>
2012/12/11
Hi,
[#50797] Re: Becoming a committer
— Charles Oliver Nutter <headius@...>
2012/12/11
I guess there's a few things I'd be interested in:
[#50809] Re: Becoming a committer
— SASADA Koichi <ko1@...>
2012/12/12
(2012/12/12 8:55), Charles Oliver Nutter wrote:
[#50815] Re: Becoming a committer
— Charles Oliver Nutter <headius@...>
2012/12/12
On Wed, Dec 12, 2012 at 12:04 AM, SASADA Koichi <ko1@atdot.net> wrote:
[#50816] Re: Becoming a committer
— "NARUSE, Yui" <naruse@...>
2012/12/12
2012/12/12 Charles Oliver Nutter <headius@headius.com>:
[#50817] Re: Becoming a committer
— Charles Oliver Nutter <headius@...>
2012/12/12
On Wed, Dec 12, 2012 at 2:59 AM, NARUSE, Yui <naruse@airemix.jp> wrote:
[#50765] [ruby-trunk - Bug #7544][Open] Accented characters in IRB — cfabianski (Cédric FABIANSKI) <cfabianski@...>
6 messages
2012/12/11
[#50793] [ruby-trunk - Bug #7547][Open] Dir.mktmpdir('~something') tries to expand a profile directory — "jstanley0 (Jeremy Stanley)" <jeremy@...>
4 messages
2012/12/11
[#50810] [ruby-trunk - Feature #7549][Open] A Ruby Design Process — "brixen (Brian Ford)" <brixen@...>
34 messages
2012/12/12
[#50829] [ruby-trunk - Feature #7549] A Ruby Design Process
— "subwindow (Erik Peterson)" <erik@...>
2012/12/12
[#50837] [ruby-trunk - Feature #7549] A Ruby Design Process
— "subwindow (Erik Peterson)" <erik@...>
2012/12/12
[#50867] [ruby-trunk - Bug #7556][Assigned] test error on refinement — "usa (Usaku NAKAMURA)" <usa@...>
14 messages
2012/12/13
[#50900] [ruby-trunk - Bug #7564][Open] r38175 introduces incompatibility — "tenderlovemaking (Aaron Patterson)" <aaron@...>
14 messages
2012/12/14
[#50913] [ruby-trunk - Bug #7568][Open] Yaml fails to encode zero date string. — "anshul (Anshul Khandelwal)" <anshul@...>
5 messages
2012/12/15
[#50920] [ruby-trunk - Bug #7568][Assigned] Yaml fails to encode zero date string.
— "charliesome (Charlie Somerville)" <charlie@...>
2012/12/16
[#50988] Re: [ruby-trunk - Bug #7568][Assigned] Yaml fails to encode zero date string.
— Aaron Patterson <tenderlove@...>
2012/12/19
On Sun, Dec 16, 2012 at 12:53:14PM +0900, charliesome (Charlie Somerville) wrote:
[#51015] 1.9.3 patch level release — Zachary Scott <zachary@...>
I know unak-san was asking about a 1.9.3 patch level release, so I
8 messages
2012/12/20
[#51099] [ruby-trunk - Feature #7612][Open] Enumerators take a proc — "pedz (Perry Smith)" <pedz@...>
4 messages
2012/12/23
[ruby-core:50584] [ruby-trunk - Feature #7472][Rejected] Add a mechanism to remove objects from the GC cycle
From:
"mame (Yusuke Endoh)" <mame@...>
Date:
2012-12-05 16:03:54 UTC
List:
ruby-core #50584
Issue #7472 has been updated by mame (Yusuke Endoh).
Status changed from Open to Rejected
----------------------------------------
Feature #7472: Add a mechanism to remove objects from the GC cycle
https://bugs.ruby-lang.org/issues/7472#change-34425
Author: sam.saffron (Sam Saffron)
Status: Rejected
Priority: Normal
Assignee:
Category:
Target version: 2.0.0
For typical rails apps there is a huge largely static object graph in memory. Requests come in, some objects are added, and the GC runs.
The blocking nature of the GC introduces a jitter that stops all execution.
In our typicalish Rails application, we see approximately 500k objects in the object space, each request add about 150k objects. On our production machine this means the GC stops execution for 30-40ms on a fairly beefy box:
GC Profiler ran during this request, if it fired you will see the cost below:
GC 457 invokes.
Index Invoke Time(sec) Use Size(byte) Total Size(byte) Total Object GC Time(ms)
1 11.169 13134480 17014400 425360 40.00200000000120326149
2 11.249 13350000 17014400 425360 32.00199999999853162080
It is clear that introducing a full scoped generational GC is a massive undertaking, but I was thinking that certain GC apis can make application servers like thin,unicorn or puma operate way more efficiently.
How about we add.
GC.skip(object) # remove object from GC cycle, essentially copy the ref to a place that is not scanned
GC.queue(object) # add an object back to the GC cycle
GC.skipped # list all the objects that are skipped
Careful use of these methods can reduce GC code by a very large amount as it would reduce fragmentation and logic required to determine if an object is alive or not.
Application servers could keep snapshots of object ids on regular intervals and decide which ones to skip. Sure memory usage would rise, but GC blocking would decrease.
Thoughts?
--
http://bugs.ruby-lang.org/