On Wed, Oct 20, 2010 at 11:14:31AM -0300, Emilio Tagua wrote:
> Hello everyone,
> 
> This email is about Marcin Raczkowski's project for Ruby Summer Of
> Code (http://rubysoc.org/projects): Project #12: ActiveRecord Identity
> Map
> 
> You can find the current implementation in my Rails repo (identity_map
> branch): http://github.com/miloops/rails/commits/identity_map
> 
> I was assigned as the mentor project and I've also invested time
> working on it, since I find its very important for all of us using
> Rails to have this implemented.
> 
> Several days ago, and after lots of feedback, I sent a pull request to
> rails/master as José Valim suggested:
> http://github.com/rails/rails/pull/76
> 
> Some key features that were lately implemented:
> 
> - A middleware that wraps each request into a new map, if it's
> disabled then no middleware is loaded.
> - IM is enabled by default: to disable it use
> config.active_record.identity_map = false
> - IM is flushed on tests setup, this was added to
> ActiveSupport::TestCase, if you are using Rspec, you should try with:
> 
> before(:each) do
>   ActiveRecord::IdentityMap.clear
> end
> 
> - A WeakHash implementation was included in ActiveSupport, if you are
> using jRuby Weakling library will be used as WeakHash.
> 
> I've been using a script to test resources, which is the most
> important feature that IM will bring among objects consistency,
> although speed has increased in some cases is not the goal of IM.
> 
> Results from the script are very promising:
> 
> http://gist.github.com/636470
> 
> I would love to get more feedback, if you try it feel free to comment
> on the pull request, reply this email, or contact me by email or IM at
> miloops at gmail.

More questions about 4db9dca55e3acc2c59f252eb83ecb83db5f4b81b:

  
http://github.com/miloops/rails/commit/4db9dca55e3acc2c59f252eb83ecb83db5f4b81b

activerecord/test/cases/autosave_association_test.rb is changed.  Why
was this test changed?  Does the AR api not work anymore?  Is it
changing?

-    assert [email protected]_for_destruction?
+    assert [email protected]_for_destruction?

Do I need to ask the new "target" object if it's marked for destruction?

I am worried.  Changing existing tests does not instill confidence. :-(

-- 
Aaron Patterson
http://tenderlovemaking.com/

Attachment: pgp59KxDYdqEq.pgp
Description: PGP signature

Reply via email to