Draft schedule of Ruby 1.9.3

dubstep · May 9, 2011, 5:44pm

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

First of all, I am sorry for proposing a plan to Ruby 1.9.3 so late.
I missed proposing it because of the earthquake in Japan.

Now Ruby 1.9.3 is much stable than on Feb. Its feature is getting
stable. It’s time to decide a plan to release Ruby 1.9.3.

I propose the following rough schedule.

The end of May: feature freeze. create ruby_1_9_3 branch.
The end of Jan: implementation freeze. don’t change without
permission after that.
The end of Jul or early Aug: release Ruby 1.9.3

The possible release blocker is the recent improvement of GVL. It is a
change in the core piece of Ruby and I would like to have it in Ruby
1.9.3 if possible.
I believe Kosaki-san is recently reviewing locks in YARV. Will it have
finished on the end of Jun?

What do you think about the schedule?

Yuki S. (Yugui) [email protected]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk3IC/MACgkQOXzH5JLb/AU/LACeIhX4xdLsjaujfXn5nZ/17Gac
SGIAoIfVUEubZG+n5ewnQRajKY1Kswl3
=j8Ig
-----END PGP SIGNATURE-----

Yuki_SSSYugui · May 9, 2011, 6:08pm

On Mon, May 9, 2011 at 12:43 PM, Yuki S. (Yugui) [email protected]
wrote:

Hi,

Hello,

The end of Jul or early Aug: release Ruby 1.9.3

The possible release blocker is the recent improvement of GVL. It is a
change in the core piece of Ruby and I would like to have it in Ruby
1.9.3 if possible.
I believe Kosaki-san is recently reviewing locks in YARV. Will it have
finished on the end of Jun?

What do you think about the schedule?

I like the schedule, as today, 32bits builds of Ruby under MinGW
generates:

ruby 1.9.3dev (2011-05-09 trunk 31496) [i386-mingw32]
9618 tests, 1880064 assertions, 22 failures, 14 errors, 83 skips

gist.github.com

https://gist.github.com/luislavena/962780

make-check.log

ruby 1.9.3dev (2011-05-09 trunk 31496) [i386-mingw32]
9618 tests, 1880064 assertions, 22 failures, 14 errors, 83 skips

  1) Skipped:
test_0002_correctly_guess_the_label_width_even_when_not_given(Benchmark::bm) [C:/Users/Luis/Projects/oss/ruby/test/benchmark/test_benchmark.rb:85]:
not_implemented

  2) Skipped:
test_sinf(DL::TestFunc) [C:/Users/Luis/Projects/oss/ruby/test/dl/test_func.rb:30]:
libm may not have sinf()

This file has been truncated. show original

There are some segfaults and stalls during the process, which turns
the automation of ruby-trunk impossible for us.

I would like to work on improve that and correct some of the false
positive tests.

1.9.3 solves some performance issues found in 1.9.2 on load and
require which had a huge impact on adoption of 1.9.2 on Windows.

Another thing that hasn’t been mentioned is the API version. It will
still be 1.9.1?

Beyond that, eagerly looking forward this release.

Yuki_SSSYugui · May 9, 2011, 6:28pm

Can you shed some light on what will happen with iconv in Ruby 1.9.3 -
will it really be gone as mentioned here:
Never use iconv by sorah · Pull Request #53 · flori/json · GitHub by sorah?

Best
Zeno

Yuki_SSSYugui · May 9, 2011, 6:47pm

The end of Jan: implementation freeze. don’t change without
permission after that.

I think you mean the end of June?

Cheers,
Rohit Arondekar
http://rohitarondekar.com

Yuki_SSSYugui · May 9, 2011, 10:43pm

On May 9, 2011, at 08:43 , Yuki S. (Yugui) wrote:

I propose the following rough schedule.

The end of May: feature freeze. create ruby_1_9_3 branch.

The end of Jan: implementation freeze. don’t change without
permission after that.

The end of Jul or early Aug: release Ruby 1.9.3

I’m a bit confused. I’m assuming that “Jan” should be “Jun” for the impl
freeze. Is that correct? If so, we’re talking about releasing 1.9.3
July/August of 2011.

There is going to be stuff for rubygems that we won’t have done until
the end of June due to personal schedules and timing.

Yuki_SSSYugui · May 9, 2011, 6:25pm

On Tue, May 10, 2011 at 12:43:46AM +0900, Yuki S. (Yugui) wrote:

finished on the end of Jun?

What do you think about the schedule?

This release schedule looks great. I am excited about the 1.9.3
release!

I am curious about the 1.9.2 patch level release though. I’ve filed
several
tickets to backport some of my commits to 1.9.2, but it seems only a few
have
been backported.

Is there a release schedule for 1.9.2? Can I help with backporting my
commits to 1.9.2?

Thank you for you hard work Yugui!

Yuki_SSSYugui · May 10, 2011, 12:41am

On May 9, 2011, at 08:43 , Yuki S. (Yugui) wrote:

The possible release blocker is the recent improvement of GVL. It is a
change in the core piece of Ruby and I would like to have it in Ruby
1.9.3 if possible.
I believe Kosaki-san is recently reviewing locks in YARV. Will it have
finished on the end of Jun?

I think this is a showstopper and needs to be addressed asap:

Xavier’s emails have gone all but ignored.

Yuki_SSSYugui · May 10, 2011, 2:40am

Hi

First of all, I am sorry for proposing a plan to Ruby 1.9.3 so late.
I missed proposing it because of the earthquake in Japan.

Now Ruby 1.9.3 is much stable than on Feb. Its feature is getting
stable. It’s time to decide a plan to release Ruby 1.9.3.

I propose the following rough schedule.

The end of May: feature freeze. create ruby_1_9_3 branch.

The end of Jan: implementation freeze. don’t change without

                    Jun?

permission after that.

The end of Jul or early Aug: release Ruby 1.9.3

The possible release blocker is the recent improvement of GVL. It is a
change in the core piece of Ruby and I would like to have it in Ruby
1.9.3 if possible.
I believe Kosaki-san is recently reviewing locks in YARV. Will it have
finished on the end of Jun?

Sure. At least, I believe so.

Yuki_SSSYugui · May 10, 2011, 2:45am

The possible release blocker is the recent improvement of GVL. It is a
change in the core piece of Ruby and I would like to have it in Ruby
1.9.3 if possible.
I believe Kosaki-san is recently reviewing locks in YARV. Will it have
finished on the end of Jun?

Sure. At least, I believe so.

$B$0$X!#2q<R$N%“%I%l%9$O(Bruby-core$B$KEPO?$7$F$J$$$NK:$l$F$F%a!<%k%j%W%i%$$7$F$7$^$C$?!#(B
$B$3$N%a!<%k$O(Bruby-core$B$KFO$$$F$$$J$$$N$G2H$K5”$C$F$+$iEj$2D>$7$^$9(B
orz

$B$H$3$m$G!"(B

Note that the mail should not be encoded in base64.

$B$@$($b$s$5$s$,%(%i!<%a%C%;!<%8$G(B [email protected]
$B$,(Bblah blah blah $B$H(B
$B$*$C$7$c$i$l$F$$$k$N$O8mM6F3$G$O$J$$$+$H;W$&$N$G$9$,!"$3$&$$$&$N$ND{@5$O(B
$BC/$K$*4j$$$9$l$P$$$$$s$G$7$g$&$M!)(B

Yuki_SSSYugui · May 10, 2011, 6:20am

$BA0ED$G$9!#(B

2011$BG/(B5$B7n(B10$BF|(B9:44 KOSAKI Motohiro
[email protected]:

$B$H$3$m$G!“(B
(snip)
$B$@$($b$s$5$s$,%(%i!<%a%C%;!<%8$G(B [email protected] $B$,(Bblah blah
blah $B$H(B
$B$*$C$7$c$i$l$F$$$k$N$O8mM6F3$G$O$J$$$+$H;W$&$N$G$9$,!”$3$&$$$&$N$ND{@5$O(B
$BC/$K$*4j$$$9$l$P$$$$$s$G$7$g$&$M!)(B

$BD>$7$H$-$^$7$?!#(B

Yuki_SSSYugui · May 10, 2011, 7:10am

$BA0ED$G$9!#(B

2011$BG/(B5$B7n(B10$BF|(B9:44 KOSAKI Motohiro [email protected]:

$B$H$3$m$G!“(B
(snip)
$B$@$($b$s$5$s$,%(%i!<%a%C%;!<%8$G(B [email protected] $B$,(Bblah blah
blah $B$H(B
$B$*$C$7$c$i$l$F$$$k$N$O8mM6F3$G$O$J$$$+$H;W$&$N$G$9$,!”$3$&$$$&$N$ND{@5$O(B
$BC/$K$*4j$$$9$l$P$$$$$s$G$7$g$&$M!)(B

$BD>$7$H$-$^$7$?!#(B

$BBP1~$"$j$,$H$&$4$6$$$^$9!*(B

Yuki_SSSYugui · May 10, 2011, 5:50am

Hello,

In message “[ruby-dev:43502] draft schedule of Ruby 1.9.3”
on May.10,2011 00:43:46, [email protected] wrote:

What do you think about the schedule?

I think current trunk is too broken, but we can repair until
the end of Jun, maybe.

BTW, I’m planning to introduce a binary incompatibility into
ruby 1.9.3. (see [ruby-dev:43450])
Please answer Go or No Go, yugui-san.

Regards

Yuki_SSSYugui · May 10, 2011, 10:50am

On Mon, May 9, 2011 at 5:43 PM, Yuki S. (Yugui) [email protected]
wrote:

Now Ruby 1.9.3 is much stable than on Feb. Its feature is getting
stable. It’s time to decide a plan to release Ruby 1.9.3.

I propose the following rough schedule.

The end of May: feature freeze. create ruby_1_9_3 branch.

The end of Jan: implementation freeze. don’t change without
permission after that.

The end of Jul or early Aug: release Ruby 1.9.3

Great news, thanks for that. It sounds like a good plan, I’m sure
we’ll be all very excited.

I wonder if there is still time for the feature proposal regarding
IPAddr and IPAddress [1] for Ruby 1.9.3.

Looks like nothing has been done so far: is there anything I can do to
help the migration, beside keep improving the library itself?

Best regards,
Marco

[1] Feature #4569: Replace IPAddr with IPAddress - Ruby master - Ruby Issue Tracking System

Yuki_SSSYugui · May 10, 2011, 12:08pm

(ruby-core$B$O$:$7$^$7$?!K(B

2011/5/10 U.Nakamura [email protected]:

Hello,

In message “[ruby-dev:43502] draft schedule of Ruby 1.9.3”
on May.10,2011 00:43:46, [email protected] wrote:

What do you think about the schedule?

I think current trunk is too broken, but we can repair until
the end of Jun, maybe.

too broken$B$H$$$o$l$k$H5$$K$J$k$N$H!"0lHV2u$7$F$=$&$J$N$OKM$@$C$?(B
$B$N$G$A$g$$$H3NG’$7$F$$^$7$?!#(B8F2E
$B$;$C$+$/$J$N$G$A$g$$$HJ,N`$7$F$$^$9!#(B

Failure:
test_binary(TestIO_M17N) [c:/ruby/trunk/test/ruby/test_io_m17n.rb:1728]:
<“a\nb\rc\r\nd\n”> expected but was
<“a\nb\nc\nd\n”>.

$B$7$P$i$/A0$K!"OCBj$K$J$C$F$$$?9TKvLdBj$N0l4S$H;W$o$l$k!#$J$+$@$5$s(B

Failure:
test_read_newline_conversion_with_encoding_conversion(TestIO_M17N)
[c:/ruby/trunk/test/ruby/test_io_m17n.rb:1505]:
<“a\r\nb\r\n”> expected but was
<“a\nb\n”>.

$B$J$+$@$5$s(B

Failure:
test_unknown_encoding(TestIconv::Basic)
[c:/ruby/trunk/test/iconv/test_basic.rb:54]:
[Iconv::InvalidEncoding] exception expected, not
Class: Iconv::BrokenLibrary
Message: <“iconv("utf-8", "X-UKNOWN")”>
—Backtrace—
c:/ruby/trunk/test/iconv/test_basic.rb:54:in iconv' c:/ruby/trunk/test/iconv/test_basic.rb:54:in block in
test_unknown_encoding’

$B$3$l$OKM$,(Bwindows $BHG(B
libiconv$B$r$A$c$s$H%S%k%I$G$-$F$J$$$@$1$J$N$GL5;k$7$F$/$@$5$$(B

Error:
test_capture3_flip(TestOpen3):
Errno::EACCES: Permission denied -
C:/Users/kosaki/AppData/Local/Temp/ruby-test20110510-3532-4090z8
c:/ruby/trunk/build/.ext/common/win32ole.rb:13:in call' c:/ruby/trunk/build/.ext/common/win32ole.rb:13:in block in
initialize’

?
$B$J$K$3$l(B

Failure:
test_commandline(TestOpen3) [c:/ruby/trunk/test/test_open3.rb:64]:
<“quux\n”> expected but was
<“quux\r\n”>.

$B$J$+$@$5$s(B

Failure:
test_include_file(TestRDocMarkupPreProcess)
[c:/ruby/trunk/test/rdoc/test_rdoc_markup_pre_process.rb:48]:
Expected “Regular expressions (regexps) are patterns which
describe the\r\ncontents of a string.\r\n”, not “Regul
ar expressions (regexps) are patterns which describe
the\ncontents of a string.\n”.

$B$J$+$@$5$s(B

Failure:
test_include_file_encoding_incompatible(TestRDocMarkupPreProcess)
[c:/ruby/trunk/test/rdoc/test_rdoc_markup_pre_process.
rb:72]:
Expected “?\r\n”, not “?\n”.

$B$J$+$@$5$s(B

Failure:
test_require_with_unc(TestRequire)
[c:/ruby/trunk/test/ruby/test_require.rb:95]:
Failed assertion, no message given.

$B$3$l$OKM$N4D6-8GM-$NLdBj$@$H@N$-$$$?5-21$,$"$k(B

Error:
test_abort_on_exception(TestThread):
IOError: stream closed
c:/ruby/trunk/test/ruby/envutil.rb:60:in read' c:/ruby/trunk/test/ruby/envutil.rb:60:in block in invoke_ruby’
c:/ruby/trunk/build/.ext/common/win32ole.rb:13:in call' c:/ruby/trunk/build/.ext/common/win32ole.rb:13:in block in
initialize’

rb_thread_io_blocking_region()$BLdBj$K$7$+8+$($J$$$N$G$J$+$@$5$s(B

Failure:
test_wait_for_invalid_fd(TestWaitForSingleFD)
[c:/ruby/trunk/test/-ext-/wait_for_single_fd/test_wait_for_single_fd.rb:27
]:
Errno::EBADF expected but nothing was raised.

$B$3$l$O!“KM$,:G6a$$$l$?(BEric$B$N(B io/wait
$B%F%9%H$J$s$G$9$,!”(BWindows$B$G$N@5$7$$F0:n$,ITL@$J$N$G(B usa
$B$5$s$N8+2r$,M_$7$$!#(B

9496 tests, 1877750 assertions, 8 failures, 2 errors, 85 skips
NMAKE : fatal error U1077: ‘.\ruby.exe’ : $B%j%?!<%s(B $B%3!<%I(B ‘0xa’
Stop.

$B0J30$HD>$;$=$&$J5$$,$7$F$-$?!#(B

Yuki_SSSYugui · May 10, 2011, 7:03pm

Hi Yugui, is there any plans for the next patch release of 1.9.2?

Currently, Ruby support is broken in Debian unstable, for instance. It
would be great to include a new patch release as soon as possible
containing this fix:

For those unfamiliar with the problem, here is a summary.

Debian has disabled SSLv2 in their last package libssl1.0.0 due to
security vulnerabilities (Ruby can be installed with 0.9.8 which was
compiled with SSLv2 enabled).

Ruby 1.9.2-p180 does not provide an option for compiling it without
SSLv2 support. RVM allows compiling openssl from source and using it
with:

http://beginrescueend.com/packages/openssl/

This has at least two problems:

1 - People concerned about security might not want to enable SSLv2;
2 - That is what most concerns me. It gets almost impossible to use the
pg gem in Debian unstable:

That is due libpq depending on system libssl which is not compatible
with the openssl provided by RVM which Ruby has been compiled against.
This results in segmentation fault keeping it very hard to have, for
instance, a Rails 3 application with PostgreSQL databases in Debian
unstable.

If #4556 is included in the next patch release to 1.9.2 this would be
much simpler to deal with.

Is there any schedule for that?

Best regards,

Rodrigo.

Em 09-05-2011 12:43, Yuki S. (Yugui) escreveu:

Yuki_SSSYugui · May 11, 2011, 4:10am

On 10/05/11 8:40 AM, Ryan D. wrote:

Skitch | Evernote
for further consideration, 1.9.3 is twice as bad again as 1.9.2:
Skitch | Evernote

I plan to look into this further on the weekend.

Cheers,
Xavier

Yuki_SSSYugui · May 10, 2011, 12:09pm

Hi

The possible release blocker is the recent improvement of GVL. It is a
change in the core piece of Ruby and I would like to have it in Ruby
1.9.3 if possible.
I believe Kosaki-san is recently reviewing locks in YARV. Will it have
finished on the end of Jun?

Sure. At least, I believe so.

Yuki_SSSYugui · May 11, 2011, 10:55am

$B$3$s$K$A$O!"$J$+$`$i(B($B$&(B)$B$G$9!#(B

In message “[ruby-dev:43513] Re: draft schedule of Ruby 1.9.3”
on May.10,2011 19:07:57, [email protected] wrote:

too broken$B$H$$$o$l$k$H5$$K$J$k$N$H!"0lHV2u$7$F$=$&$J$N$OKM$@$C$?(B
$B$N$G$A$g$$$H3NG’$7$F$_$^$7$?!#(B8F2E

$B$A$J$$K!“;d$N4D6-$O(Bfiddle$B$,;H$($k$N$G$9$,!”$=$$$D$,:G6a(BE$B$@$i(B
$B$1$K$J$C$F$F!"$=$l$r4^$a$F:#F|8=:$G9g7W(B6F16E$B$G$9$M!#(B
$B$H$j$“$($:!”(BE$B$,(B2$B7e$H$+$"$C$?$i(Btoo
broken$B$H8@$o$l$F$b$7$c!<$J(B
$B$$$H;d$O;W$$$^$9!#(B

$B$&$A$N4D6-0MB8$+$b$7$l$J$$$N$G!"??LLL$KD4::$7$F$+$iJs9p!&(B

$BBP1~$NM=Dj$G$O$"$j$^$9!#(B

$B$=$l$G$O!#(B

Yuki_SSSYugui · May 11, 2011, 9:02am

On 11/05/11 at 02:03 +0900, Rodrigo Rosenfeld R. wrote:

Hi Yugui, is there any plans for the next patch release of 1.9.2?

Currently, Ruby support is broken in Debian unstable, for instance.
It would be great to include a new patch release as soon as possible
containing this fix:

Feature #4556: Please add option which enable SSLv2 support - Ruby master - Ruby Issue Tracking System

For clarification, the Ruby 1.9.2 package in Debian has already been
fixed
by backporting a patch.

Lucas

Yuki_SSSYugui · May 11, 2011, 12:39pm

$B=uED$G$9!#(B

On Tue, May 10, 2011 at 07:07:57PM +0900, KOSAKI Motohiro wrote:

Error:
test_capture3_flip(TestOpen3):
Errno::EACCES: Permission denied -
C:/Users/kosaki/AppData/Local/Temp/ruby-test20110510-3532-4090z8
c:/ruby/trunk/build/.ext/common/win32ole.rb:13:in call' c:/ruby/trunk/build/.ext/common/win32ole.rb:13:inblock in initialize’

?
$B$J$K$3$l(B

$B<j85$G%F%9%H$,$G$-$F$$$J$$$N$H!"(Btest_capture3_flip
$B$,2?$N%F%9%H$r$7$F$$$k$N$+(B
$BCN$i$J$$$N$G!“3’L\8!F$$,$D$+$J$$$N$G$9$,(B…
win32ole.rb$BB&$G!”(BThread#initialize $B$r:FDj5A$7$F$$$^$9!#(B
$B$=$l$,2?$+0-$5$7$F$^$9$+$M!)(B

$B=uED(B $B2m5*(B