[#35961] require performance on 1.9 — Xavier Shay <xavier-list@...>
Hello,
[#35985] [Backport92 - Backport #4641][Open] Please backport r31418 to 1.9.2 stable branch — Aaron Patterson <aaron@...>
[#36013] [Ruby 1.9 - RubySpec #4649][Open] Adding parallel constructors to Ruby 2.0 — Rodrigo Rosenfeld Rosas <rr.rosas@...>
[#36046] [Ruby 1.9 - Bug #4655][Open] String#to_c does not support scientific notation — Tinco Andringa <mail@...>
[#36058] draft schedule of Ruby 1.9.3 — "Yuki Sonoda (Yugui)" <yugui@...>
-----BEGIN PGP SIGNED MESSAGE-----
Hi Yugui, is there any plans for the next patch release of 1.9.2?
[#36108] [Ruby 1.9 - Bug #4666][Open] set ruby compatibility version to 1.9.3 in trunk — Lucas Nussbaum <lucas@...>
Lucas Nussbaum <lucas@lucas-nussbaum.net> wrote:
> Even if 1.9.3 is still binary-compatible with 1.9.1, I think that it would be easier to change
2011/5/12 Urabe Shyouhei <shyouhei@ruby-lang.org>:
[#36131] Re: [ruby-cvs:38172] Ruby:r30989 (trunk): * include/ruby/win32.h: define WIN32 if neither _WIN64 nor WIN32 defined. it forces to use push/pop for pack(4) pragma. — "Yuki Sonoda (Yugui)" <yugui@...>
Hi arton,
Hi,
[#36150] [Ruby 1.9 - Bug #4680][Open] [PATCH] io.c: fix busy wait with sendfile() — Eric Wong <normalperson@...>
[#36156] [Ruby 1.9 - Bug #4683][Open] [PATCH] io.c: copy_stream execute interrupts and retry — Eric Wong <normalperson@...>
[#36167] [Ruby 1.9 - Bug #4421] [ext/openssl] Fix RSA public key encoding — Hiroshi NAKAMURA <nakahiro@...>
[#36255] Whitespace conventions? — Steve Klabnik <steve@...>
So, while working on some documentation, I've noticed that there's a lot of
2011/5/17 Steve Klabnik <steve@steveklabnik.com>:
[#36285] unable to load irb, 1.9.3 mingw — Roger Pack <rogerdpack2@...>
Hello all, with mingw 1.9.3, I get the following when trying to load irb:
[#36314] [Ruby 1.9 - Bug #3167] RDoc issues in interactive mode — Benoit Daloze <redmine@...>
[#36316] [Ruby 1.9 - Bug #4731][Open] ruby -S irb fails with mingw/msys vanilla builds — Roger Pack <rogerpack2005@...>
Hi,
On Sat, Jul 16, 2011 at 7:46 AM, Nobuyoshi Nakada <nobu@ruby-lang.org> wrote:
[#36322] [Ruby 1.9 - Bug #4734][Assigned] [ext/openssl] DSA#sign error — Martin Bosslet <Martin.Bosslet@...>
[#36337] [Ruby 1.9 - Feature #3905] rb_clear_cache_by_class() called often during GC for non-blocking I/O — Akira Tanaka <akr@...>
Akira Tanaka <akr@fsij.org> wrote:
Hi,
SASADA Koichi <ko1@atdot.net> wrote:
[#36373] [Ruby 1.9 - Bug #4757][Open] Attempt to make Enumerator docs more clear (patch included) — David Copeland <davetron5000@...>
2011/5/25 Yusuke Endoh <mame@tsg.ne.jp>:
[#36374] [Ruby 1.9 - Bug #4758][Open] yaml file not human readable when saving utf-8 — Ilias Lazaridis <ilias@...>
[#36390] [Ruby 1.9 - Feature #4766][Open] Range#bsearch — Yusuke Endoh <mame@...>
On Jul 17, 2011, at 7:54 PM, Eric Hodel wrote:
[#36395] [Ruby 1.9 - Bug #4769][Open] Updated SMTP standards — "J.R. Garcia" <mrjohngarcia@...>
[#36406] 1.8.7 release next month — Urabe Shyouhei <shyouhei@...>
Hello core people,
2011/5/23 Urabe Shyouhei <shyouhei@ruby-lang.org>:
Hi Luis,
From: Urabe Shyouhei <shyouhei@ruby-lang.org>
From: Hidetoshi NAGAI <nagai@ai.kyutech.ac.jp>
Ping Luis, how's it going?
On Fri, Jun 3, 2011 at 5:18 AM, Urabe Shyouhei <shyouhei@ruby-lang.org> wrote:
Hi,
On Sun, Jun 5, 2011 at 4:30 AM, Hidetoshi NAGAI <nagai@ai.kyutech.ac.jp> wrote:
(06/06/2011 01:16 PM), Luis Lavena wrote:
On Mon, Jun 6, 2011 at 1:24 AM, Urabe Shyouhei <shyouhei@ruby-lang.org> wrote:
From: Luis Lavena <luislavena@gmail.com>
[#36419] [Ruby 1.9 - Feature #4772][Open] Hash#add_keys — Joey Zhou <yimutang@...>
[#36429] GC thought — Roger Pack <rogerdpack2@...>
Hello all.
[#36447] [Ruby 1.9 - Bug #4777][Open] Ruby 1.9.2-p180 ignoring INT, TERM, and QUIT until it receives CONT — Nathan Sobo <nathansobo@...>
[#36463] [Ruby 1.9 - Feature #4778][Open] IO#each_chomped — Joey Zhou <yimutang@...>
[#36474] Error reporting, backtraces and the debugger — Clifford Heath <clifford.heath@...>
Dear people,
[#36479] [Ruby 1.9 - Feature #4784][Open] Import the JSON library — Lazaridis Ilias <ilias@...>
[#36494] [Ruby 1.9 - Feature #4786][Open] RCR new Feature: Numeric#grouped — Roger Pack <rogerpack2005@...>
[#36528] [Ruby 1.9 - Bug #4795][Open] Nested classes don't seem to resolve correctly when another class exists with the same name — John Feminella <johnf@...>
[#36536] [Ruby 1.9 - Bug #3924] Performance bug (in require?) — Xavier Shay <xavier-list@...>
[#36550] [Ruby 1.9 - Bug #4798][Open] test_process and test_signal errors and halts on Windows — Luis Lavena <luislavena@...>
[#36551] [Ruby 1.9 - Bug #4799][Open] M17N tests are too JP specific — Luis Lavena <luislavena@...>
[#36558] [Ruby 1.9 - Bug #3924] Performance bug (in require?) — Xavier Shay <xavier-list@...>
Hello,
Hello, Xavier
[#36559] [Ruby 1.9 - Feature #4801][Open] Shorthand Hash Syntax for Strings — Tom Wardrop <tom@...>
Hi,
> Iff 'key': 'value'} means {:key => 'value'} I have no objection.
Hi,
On Mon, May 30, 2011 at 04:21:32PM +0900, Yukihiro Matsumoto wrote:
Em 30-05-2011 07:58, Cezary escreveu:
Since :"#{abc}" is allowed in Ruby, I imagine that any such substitute syntax would preserve that property.
Em 30-05-2011 09:05, Michael Edgar escreveu:
On Mon, May 30, 2011 at 09:05:04PM +0900, Michael Edgar wrote:
Cezary:
On Tue, May 31, 2011 at 05:55:39AM +0900, Piotr Szotkowski wrote:
On May 30, 2011, at 10:19 AM, Cezary wrote:
On 5/30/11 9:24 AM, Michael Edgar wrote:
On 02/06/2011, at 10:28 AM, Kurt Stephens wrote:
On 6/1/11 10:17 PM, Bill Kelly wrote:
[#36565] [Ruby 1.9 - Bug #4803][Open] RCLASS_SUPER won't compile for C extensions as of revision 31627 — Daniel Azuma <dazuma@...>
Hi,
[#36628] [Ruby 1.9 - Feature #4805][Open] Add X509::Name#hash_old for 0.9.X compat — Hiroshi NAKAMURA <nakahiro@...>
[ruby-core:36308] [Ruby - Bug #921][Open] autoload is not thread-safe
Issue #921 has been updated by Hiroshi NAKAMURA.
Status changed from Closed to Open
Assignee changed from Nobuyoshi Nakada to Shyouhei Urabe
ruby -v changed from - to ruby 1.8.7 (2011-02-18 patchlevel 334) [x86_64-linux]
For JRuby 1.9, I fixed this issue (autoload thread safety.)
Regards to autoload, CRuby 1.9 is thread-safe. CRuby 1.8 is not thread-safe.
% cat autoload.rb
class Foo
autoload :X, 'constant.rb'
end
Thread.abort_on_exception = true
t1 = Thread.new {
puts "thread #{Thread.current} accessing X"
p Foo::X
}
t2 = Thread.new {
puts "thread #{Thread.current} accessing X"
p Foo::X
}
t1.join
t2.join
% cat constant.rb
# simulate a slow file load or a deep chain of requires
puts "#{Thread.current} in constant.rb"
1_000_000.times { Thread.pass }
class Foo
# define X
X = 1
end
% ruby187 autoload.rb
thread #<Thread:0x7f2e2e301de0> accessing X
#<Thread:0x7f2e2e301de0> in constant.rb
thread #<Thread:0x7f2e2e301390> accessing X
autoload.rb:13: uninitialized constant Foo::X (NameError)
from autoload.rb:11:in `initialize'
from autoload.rb:11:in `new'
from autoload.rb:11
zsh: exit 1 ruby187 autoload.rb
After talking to Shyouhei at Asakusa.rb meetup last night, we agreed that
it's a bug and it should be fixed if we can.
And here's another problem arises. 1.8 removes autoload a constant first as
Charles stated above, then requires the specified file. It's the cause of
threaded autoload issue but it has another side effect. When the require fails
with some Exception, the defined constant is removed afterwards.
% ruby187 -I. -e 'autoload(:X, "X"); begin; X; rescue LoadError; end; p Object.constants.include?("X")'
false
1.9 does not remove the constant.
% ruby -I. -e 'autoload(:X, "X"); begin; X; rescue LoadError; end; p Object.constants.include?(:X)'
true
I'm guessing this behavior would be affected when we fix 1.8's thread safety.
Unfortunately, there're "spec"s in RubySpec which expects that constant to be
removed in 1.8.
Can we change this 1.8.7 behavior?
----------------------------------------
Bug #921: autoload is not thread-safe
http://redmine.ruby-lang.org/issues/921
Author: Charles Nutter
Status: Open
Priority: Normal
Assignee: Shyouhei Urabe
Category:
Target version:
ruby -v: ruby 1.8.7 (2011-02-18 patchlevel 334) [x86_64-linux]
=begin
Currently autoload is not safe to use in a multi-threaded application. To put it more bluntly, it's broken.
The current logic for autoload is as follows:
1. A special object is inserted into the target constant table, used as a marker for autoloading
2. When that constant is looked up, the marker is found and triggers autoloading
3. The marker is first removed, so the constant now appears to be undefined if retrieved concurrently
4. The associated autoload resource is required, and presumably redefines the constant in question
5. The constant lookup, upon completion of autoload, looks up the constant again and either returns its new value or proceeds with normal constant resolution
The problem arises when two or more threads try to access the constant. Because autoload is stateful and unsynchronized, the second thread may encounter the constant table in any number of states:
1. It may see the autoload has not yet fired, if the first thread has encountered the marker but not yet removed it. It would then proceed along the same autoload path, requiring the same file a second time.
2. It may not find an autoload marker, and assume the constant does not exist.
3. It may see the eventual constant the autoload was intended to define.
Of these combinations, (3) is obviously the desired behavior. (1) can only happen on native-threaded implementations that do not have a global interpreter lock, since it requires concurrency during autoload's internal logic. (2) can happen on any implementation, since while the required file is processing the original autoload constant appears to be undefined.
I have only come up with two solutions:
* When the autoload marker is encountered, it is replaced (under lock) with an "autoload in progress" marker. All subsequent threads will then see this marker and wait for the autoloading process to complete. the mechanics of this are a little tricky, but it would guarantee concurrent autoloads would only load the target file once and would always return the intended value to concurrent readers.
* A single autoload mutex, forcing all autoloads to happen in serial.
There is a potential for deadlock in the first solution, unfortunately, since two threads autoloading two constants with circular autoloaded constant dependencies would ultimately deadlock, each waiting for the other to complete. Because of this, a single autoload mutex for all autoloads may be the only safe solution.
=end
--
http://redmine.ruby-lang.org