Last call for patches - 2.3 final is imminent

Pratik

unread,

Mar 9, 2009, 8:03:26 AM3/9/09

to rubyonra...@googlegroups.com

Hey all,

As 2.3 RC2 is out, the plan is to have the 2.3 final release out as
early as possible. So we'd like to make the last call for patches. If
you have any pending patches that you want in 2.3 final release,
please speak now. Not very keen on accepting refactoring patches or
new features at this point as they can wait. But if there are any
critical bug fixes or anything that should block the release, do post
it in the replies.

Thanks for all the contributions. 2.3 is looking really solid !

--
Cheers!
- Pratik
http://m.onkey.org

Johan Sørensen

unread,

Mar 9, 2009, 10:56:02 AM3/9/09

to rubyonra...@googlegroups.com

On Mon, Mar 9, 2009 at 1:03 PM, Pratik <prati...@gmail.com> wrote:
>
> Hey all,
>
> As 2.3 RC2 is out, the plan is to have the 2.3 final release out as
> early as possible. So we'd like to make the last call for patches. If
> you have any pending patches that you want in 2.3 final release,
> please speak now. Not very keen on accepting refactoring patches or
> new features at this point as they can wait. But if there are any
> critical bug fixes or anything that should block the release, do post
> it in the replies.

The only blocker for me is a Ruby 1.9 issue:
http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/1988

As I wrote in the ticket, this seems to be the only place this is
needed due to the nature of the partials+content_for blocks, so magic
comments won't work here.

> --
> Cheers!
> - Pratik
> http://m.onkey.org

JS

Andrew Bloom

unread,

Mar 9, 2009, 11:21:33 AM3/9/09

to Ruby on Rails: Core

This isn't a show stopper or a blocker, but it's really simple.
http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2122-routing-with-a-blank-path_prefix-causes-a-bad-url-to-generate

Any chance we could get it accepted?

Thanks
Andrew

On Mar 9, 9:56 am, Johan Sørensen <jo...@johansorensen.com> wrote:

David Genord II

unread,

Mar 9, 2009, 11:44:19 AM3/9/09

to rubyonra...@googlegroups.com

Also not a show stopper but it would e nice to get some eyes on this one.
http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2179-activeresource-not-eager-loading-namespaced-models

David Genord II

Sam Granieri

unread,

Mar 9, 2009, 12:31:03 PM3/9/09

to Ruby on Rails: Core

I'd like to get some eyes on this one as well
http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2181-default-scope-with-conditions-on-create

Milhouse

unread,

Mar 10, 2009, 8:32:43 AM3/10/09

to Ruby on Rails: Core

Not a critical blocker, but the first patch in :
http://rails.lighthouseapp.com/projects/8994/tickets/2000-patch-for-http-digest-authentication-uri-comparison#ticket-2000-11

has an easy correction to the earlier commit. Other issues re HTTP
Digest auth can wait, although I believe it is broken when the
connection is sessionless, like from cURL without cookies turned on.

Thanks,

Don

fowlduck

unread,

Mar 10, 2009, 10:23:32 AM3/10/09

to Ruby on Rails: Core

I'd say this is critical, but I don't know who else would agree:
http://rails.lighthouseapp.com/projects/8994/tickets/2115-malformed-xml-request-returns-vague-error

Leonardo Borges

unread,

Mar 10, 2009, 10:25:47 AM3/10/09

to Ruby on Rails: Core

This would be really helpful for us here:
http://rails.lighthouseapp.com/projects/8994/tickets/1874-add-short-hand-definition-for-postgresql-xml-data-type

Tks

On Mar 10, 1:32 pm, Milhouse <donald.par...@gmail.com> wrote:
> Not a critical blocker, but the first patch in :

> http://rails.lighthouseapp.com/projects/8994/tickets/2000-patch-for-h...

David Dollar

unread,

Mar 10, 2009, 10:56:58 AM3/10/09

to Ruby on Rails: Core

I have a patch that fixes a lot of the fragility in the current gems:*
rake tasks, as well as making unpacking a lot more stable.

Please check it out at:

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2195-patch-fix-the-gem-dependency-system-in-rails#ticket-2195-2

The current rake gems tasks do not work in many configurations.

Chris Cruft

unread,

Mar 10, 2009, 12:00:21 PM3/10/09

to Ruby on Rails: Core

Not a patch, but last I checked assert_select_email didn't work -
undefined method in ActionMailer::TestCase subclasses.

Henrik N

unread,

Mar 10, 2009, 12:19:38 PM3/10/09

to Ruby on Rails: Core

Pratik wrote:
> Hey all,
>
> As 2.3 RC2 is out, the plan is to have the 2.3 final release out as
> early as possible. So we'd like to make the last call for patches.

Minor regression, "auto_link should not linkify URLs in the middle of
a tag": http://rails.lighthouseapp.com/projects/8994/tickets/1523-auto_link-should-not-linkify-urls-in-the-middle-of-a-tag

mwhuss

unread,

Mar 10, 2009, 12:28:49 PM3/10/09

to Ruby on Rails: Core

When using rails 2.3.1 (does not apply to 2.3.0) I am having issues
with destroying sessions (or logging out). It also only applies when
using passenger.

Conditions:
Rails 2.3.1
Passenger 2.0.6
Both Restful Authentication and Clearance

When using mongrel everything works as it should.

Marshall

Bira

unread,

Mar 10, 2009, 12:56:27 PM3/10/09

to Ruby on Rails: Core

On Mar 9, 9:03 am, Pratik <pratikn...@gmail.com> wrote:
> Hey all,
>
> As 2.3 RC2 is out, the plan is to have the 2.3 final release out as
> early as possible. So we'd like to make the last call for patches. If
> you have any pending patches that you want in 2.3 final release,
> please speak now.

> --
> Cheers!
> - Pratikhttp://m.onkey.org

Hello! I just joined the group through Google. I hope I'm not
intruding, but there's something I would like to see in 2.3:

http://rails.lighthouseapp.com/projects/8994/tickets/1100-json-decoding-of-u-escape-sequences

I don't know if it can be called a blocker, since it doesn't prevent
Rails from running, but it causes significant annoyance if JSON is
used heavily in an app.
It's been extant since 2.0, I think (http://dev.rubyonrails.org/ticket/
8599) - we've been using a locally-patched Rails for about that long
where I work. Earlier today I've reformatted the patch on the
Lighthouse ticket so it applies on today's master, and it's very close
to being approved.

--
Bira
http://compexplicita.tumblr.com

James Miller

unread,

Mar 10, 2009, 1:15:05 PM3/10/09

to Ruby on Rails: Core

Marshall,

Check out Passenger 2.1.1 at
http://blog.phusion.nl/2009/03/01/phusion-passenger-211-beta-released-thanks-sponsors/

There appears to be some compatibility issues with 2.0.6.

James

Pratik

unread,

Mar 10, 2009, 1:16:42 PM3/10/09

to rubyonra...@googlegroups.com

Yup. You'll need 2.1.1 to run Rails 2.3. I just had a chat with Hongli
and he said he'll be trying his best to push the passenger release out
asap.

Marshall Huss

unread,

Mar 10, 2009, 1:18:34 PM3/10/09

to rubyonra...@googlegroups.com

I just installed passenger 2.1.1 and tested both clearance and restful-auth, it fixed the issue with both.

Marshall

--
Marshall

russ

unread,

Mar 10, 2009, 1:33:14 PM3/10/09

to Ruby on Rails: Core

This bug here stops Litespeed from working at all on RC2.

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2198-litespeed-compatibility

Ken Collins

unread,

Mar 10, 2009, 4:29:23 PM3/10/09

to Ruby on Rails: Core

I consider this Mocha compatibility very important.

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2060-installation-of-miniunit-causes-test-to-fail#ticket-2060-4

rubys

unread,

Mar 10, 2009, 6:00:50 PM3/10/09

to Ruby on Rails: Core

On Mar 9, 8:03 am, Pratik <pratikn...@gmail.com> wrote:

The depot application in the Agile Web Development with Rails, Edition
3 is broken with the current git master (it got considerably further
with 2.3rc2). Details:

http://intertwingly.net/blog/2009/03/10/Rails-System-Integration-Testing

> --
> Cheers!
> - Pratikhttp://m.onkey.org

- Sam Ruby

Mislav Marohnić

unread,

Mar 10, 2009, 7:15:09 PM3/10/09

to rubyonra...@googlegroups.com

Action caching is broken with string keys.

This means you cannot cache XML or JSON responses, for instance, if you use a string for cache key.

Failing test + fix http://rails.lighthouseapp.com/projects/8994/tickets/1585

Carlos A. da Silva

unread,

Mar 10, 2009, 8:23:00 PM3/10/09

to Ruby on Rails: Core

I've tested the new feature for processing records in batches, and I
noticed the each method does not work with named_scopes. I attempted
to make a simple patch, but added no tests yet.

I think it's a really simple change that can help developers a lot
saving time and making the code more DRY.

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2201-batches-each-with-named_scope-doesnt-seem-to-work

Please take a look.

--
Carlos A. da Silva

Ryan Bigg

unread,

Mar 11, 2009, 1:33:05 AM3/11/09

to rubyonra...@googlegroups.com

I'd like to see this fixed:

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2208-rails-231-runs-into-a-segfault
, as it concerns Ruby 1.9 and Mysql.

Milhouse

unread,

Mar 11, 2009, 3:01:08 AM3/11/09

to Ruby on Rails: Core

I've replaced this with a newer patch: http://rails.lighthouseapp.com/projects/8994/tickets/2209

Don

On Mar 10, 7:32 am, Milhouse <donald.par...@gmail.com> wrote:
> Not a critical blocker, but the first patch in :

> http://rails.lighthouseapp.com/projects/8994/tickets/2000-patch-for-h...

Xavier Noria

unread,

Mar 11, 2009, 3:27:44 AM3/11/09

to rubyonra...@googlegroups.com

More wood!

http://rails.lighthouseapp.com/projects/8994/tickets/1995-skip_relative_url_root-is-broken

jodosha

unread,

Mar 11, 2009, 7:41:18 AM3/11/09

to Ruby on Rails: Core

Not a blocker, but around I've seen to much custom solutions for
running plugin migrations.
I created a patch for this, I think it could work for the 95% of the
existing applications.

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2058

Cheers,
Luca.

Will Bryant

unread,

Mar 11, 2009, 9:17:52 AM3/11/09

to rubyonra...@googlegroups.com

There's a new regression in RC2 from the autosave logic, on any object
with a has_one :through that's been :included (even if autosave is
off). Very small patch:

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2213-saving-objects-with-associated-has_one-throughs-loaded-broken

Cheers,
Will

Jacob

unread,

Mar 11, 2009, 10:49:33 AM3/11/09

to Ruby on Rails: Core

I've been having a lot of issues with the new autosave feature in 2.3
I could easily workaround most of these issues if I had the ability to
explicitly disable autosave for certain relationships in such a way
that would override the autosave that happens when the association is
a "new_record?"

Discussion is here: http://rails.lighthouseapp.com/projects/8994/tickets/2144

I am working on a patch and will be filing a separate ticket for it
shortly.

Ryan Angilly

unread,

Mar 11, 2009, 11:02:41 AM3/11/09

to rubyonra...@googlegroups.com

I've been trying to keep up to speed on all the stuff going into 2.3, but I missed mention of "the new autosave feature". Google only returns links to lighthouse tickets that aren't very explanatory. Could someone link me to it?

Eloy Duran

unread,

Mar 11, 2009, 11:16:48 AM3/11/09

to rubyonra...@googlegroups.com

I've been having a lot of issues with the new autosave feature in 2.3
I could easily workaround most of these issues if I had the ability to
explicitly disable autosave for certain relationships in such a way
that would override the autosave that happens when the association is
a "new_record?"

You don't have any issues with the new autosave behaviour, rather with the behaviour which already existed. But we went over that already on the ticket.

Discussion is here: http://rails.lighthouseapp.com/projects/8994/tickets/2144

I am working on a patch and will be filing a separate ticket for it
shortly.

Looking forward to it.

Eloy

Eloy Duran

unread,

Mar 11, 2009, 11:17:04 AM3/11/09

to rubyonra...@googlegroups.com

Sure here you go: http://github.com/rails/rails/blob/master/activerecord/lib/active_record/autosave_association.rb

It was part of the NestedAttributes patch so it wasn't the most interesting feature from that patch I guess :)

Eloy

Ryan Angilly

unread,

Mar 11, 2009, 11:22:06 AM3/11/09

to rubyonra...@googlegroups.com

Pretty neat. Thanks

Jason King

unread,

Mar 11, 2009, 11:23:42 AM3/11/09

to rubyonra...@googlegroups.com

Just wanted to add this to the list, DDLs in sqlite3 transactions (and
savepoints and version comparison bugfixes to-boot). Koz is already
looking at it - but thought I should mention it in here in case:

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2080-sqlite-missing-ddl-transactions-and-savepoints

Murray Steele

unread,

Mar 11, 2009, 12:06:58 PM3/11/09

to rubyonra...@googlegroups.com

Like most of the one's mentioned, not a show stopper, but it's ready and waiting for a commit I think: http://rails.lighthouseapp.com/projects/8994/tickets/1696-problem-with-nested-include-missing-data-and-old-style-single-query-eager-loading

Also, understandably not likely for 2.3, but more eyes and thinking about this one to see if can get into whatever version of rails comes out next would be good: http://rails.lighthouseapp.com/projects/8994/tickets/1619-patch-support-for-inverse-option-in-associations

Cheers,

Muz

2009/3/9 Pratik <prati...@gmail.com>

Jacob Burkhart

unread,

Mar 11, 2009, 12:31:41 PM3/11/09

to rubyonra...@googlegroups.com

Hey Eloy,

Thanks for your patience with me... Here's my new ticket and patch
for autosave:
http://rails.lighthouseapp.com/projects/8994/tickets/2214

David Stevenson

unread,

Mar 11, 2009, 12:35:38 PM3/11/09

to Ruby on Rails: Core

Small, easy, well tested fixes for radio_button & label:

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/1993-fradio_button-ignores-parent-fields_fors-index-option

New patch submitted today.

Jason King

unread,

Mar 11, 2009, 1:55:55 PM3/11/09

to rubyonra...@googlegroups.com

Also, this one fixes kind-of a bug:
http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2162-layout-chaining

At the moment, if you put two layout calls with options into a
controller, then the options will be merged together, but the template
name will be last-in-wins. This can be confusing, and the docs don't
actually say not to put two layout calls in one controller, AND the
layout guide actually gives this as an example of usage.

Instead of fixing the guide, I thought it was a nice idea to have
multiple layout methods, so I wrote this patch - which also fixes the
kind-of bug of conditions being merged but layout names being
overwritten.

Josh Susser

unread,

Mar 11, 2009, 2:32:34 PM3/11/09

to rubyonra...@googlegroups.com

This seems more like a feature addition and less like a bug fix. I'd
say hold off until after the release for it.

--josh

--
Josh Susser
http://blog.hasmanythrough.com
Golden Gate Ruby Conf :: April 17-18 :: http://gogaruco.com

Pratik

unread,

Mar 11, 2009, 2:35:14 PM3/11/09

to rubyonra...@googlegroups.com

Agree with Josh. Let's refrain from "good to have" stuff and focus on
the "must have" patches.

Bira

unread,

Mar 11, 2009, 5:04:43 PM3/11/09

to Ruby on Rails: Core

On Tue, Mar 10, 2009 at 1:56 PM, Bira <u.alb...@gmail.com> wrote:

> intruding, but there's something I would like to see in 2.3:
>
> http://rails.lighthouseapp.com/projects/8994/tickets/1100-json-decoding-of-u-escape-sequences

There were some hiccups with 1.9 on this one, but they should be fixed
in the latest patch I posted :).

--
Bira
http://compexplicita.wordpress.com
http://compexplicita.tumblr.com

Jacob Burkhart

unread,

Mar 11, 2009, 6:42:47 PM3/11/09

to rubyonra...@googlegroups.com

An association reload oddity to add to the list:
http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2217-association-reload-fails-due-to-cached-finder_sql

Jason King

unread,

Mar 11, 2009, 7:34:40 PM3/11/09

to rubyonra...@googlegroups.com

See? Because of me, now they have a warning.

José Valim

unread,

Mar 12, 2009, 4:57:52 AM3/12/09

to Ruby on Rails: Core

This is a fix in Model.human_name, which returns "Postcomment" when it
should return "Post comment".

http://rails.lighthouseapp.com/projects/8994/tickets/2120-humanize-and-human_name-dont-separate-words

José Valim

unread,

Mar 12, 2009, 4:59:19 AM3/12/09

to Ruby on Rails: Core

And this one makes labels work with human_attribute_name.
It's not a blocker, but IMHO it's an I18n bug:

http://rails.lighthouseapp.com/projects/8994/tickets/745-form-label-should-use-i18n

Stevie D

unread,

Mar 12, 2009, 10:12:30 AM3/12/09

to Ruby on Rails: Core

How hard would it be to patch a naming convention in nested
attributes?

I'd be interested in patching myself, though my testing skills
probably should not be trusted.

This is basically what I'd like to see:

http://pastie.textmate.org/pastes/413750

Anyone else think this would be a marked improvement?

Eloy Duran

unread,

Mar 12, 2009, 10:31:33 AM3/12/09

to rubyonra...@googlegroups.com

It won't be hard at all, in fact it _was_ this way in the beginning.
However, after much discussion it was decided that the current way is
the better way and although it takes a bit more typing, I agree with
the decision.

Example:

“:attributes => true” doesn't say anything about the association,
which is what has_many (and the other macros) are about.
Now that example could be fixed easily, however it gets worse with the
options for NestedAttributes. “:reject_if => …” has nothing to do with
the association at all.

There were some other minor issues as wel iirc, you can search the ML
and LH for that.

To conclude; I feel nothing for changing this, and especially not when
the core team is trying to to get out the must have patches.

Cheers,
Eloy

Luke Melia

unread,

Mar 13, 2009, 5:39:29 PM3/13/09

to rubyonra...@googlegroups.com

I just posted to

http://rails.lighthouseapp.com/projects/8994-ruby-on-rails/tickets/2054-requirements-not-passed-to-named_route-generated-by-member#ticket-2054-9

The commit on that ticket (5b7527ca44521edf9782b3d7f449bf09a29267f2)
broke routes where a resource has a collection and member action with
the same name.

On Mar 10, 2009, at 10:23 AM, fowlduck wrote:

>
> I'd say this is critical, but I don't know who else would agree:
> http://rails.lighthouseapp.com/projects/8994/tickets/2115-malformed-xml-request-returns-vague-error
>
> >

--
Luke Melia
lu...@lukemelia.com
http://www.lukemelia.com/

Mike Gunderloy

unread,

Mar 13, 2009, 8:33:31 PM3/13/09

to rubyonra...@googlegroups.com

Thanks for catching that, Luke. Commit has been reverted until we come
up with a more global solution.

Mike

jodosha

unread,

Mar 15, 2009, 4:52:10 PM3/15/09

to Ruby on Rails: Core

I found a little issue in Metal file name convention:
I have a Metal piece file called legacy_routes.rb with the related
class called LegacyRoutes, when I try to start the server it returns
the following error:

/Library/Ruby/Gems/1.8/gems/activesupport-2.3.0/lib/active_support/
dependencies.rb:443:in `load_missing_constant': uninitialized constant
LegacyRoute (NameError)
# ...

I created a patch and a LH ticket: http://tinyurl.com/db9wnz

Luca
---
blog: www.lucaguidi.com

Stephen Bannasch

unread,

Mar 16, 2009, 6:20:53 PM3/16/09

to rubyonra...@googlegroups.com

>Hey all,
>
>As 2.3 RC2 is out, the plan is to have the 2.3 final release out as
>early as possible. So we'd like to make the last call for patches. If
>you have any pending patches that you want in 2.3 final release,
>please speak now. Not very keen on accepting refactoring patches or
>new features at this point as they can wait. But if there are any
>critical bug fixes or anything that should block the release, do post
>it in the replies.
>
>Thanks for all the contributions. 2.3 is looking really solid !

I'd like to see the patch in this ticket added:

Add JDOM (JRuby) as XmlMini backend
http://rails.lighthouseapp.com/projects/8994/tickets/2238

I'm working on adapting Aarons, work in
test/xml_mini/nokogiri_engine_test.rb into a patch for testing all
the alternate backends for xml_mini.

Here's the current state of the test:

http://github.com/stepheneb/rails_hash_from_xml/blob/823a8b6b4c2bad04a27e9131082d816e8c19990e/test/test_rexml_equality.rb

Right now the Nokogiri and the JDOM adapter pass all 10 tests I've adapted.

LibXML fails 2.

FYI: speed converting a 1.8MB xml file to a hash using Hash.from_xml

* libxml-ruby was 25 faster than REXML
* nokogiri was only 1.6 times faster
* JRuby-REXML was 3 times faster than MRI 1.8.6
* JRuby-JDOM was 10 times faster

Pratik

unread,

Mar 16, 2009, 6:21:54 PM3/16/09

to rubyonra...@googlegroups.com

Little too late for 2.3, don't you think ;-)
http://weblog.rubyonrails.org/2009/3/16/rails-2-3-templates-engines-rack-metal-much-more

Stephen Bannasch

unread,

Mar 16, 2009, 6:44:21 PM3/16/09

to rubyonra...@googlegroups.com

At 10:21 PM +0000 3/16/09, Pratik wrote:
>Little too late for 2.3, don't you think ;-)
>http://weblog.rubyonrails.org/2009/3/16/rails-2-3-templates-engines-rack-metal-much-more