Datamapper DataMapper

Comments and changes to this ticket

• You flagged this item as spam.
  

  Michael Boutros February 10th, 2008 @ 08:22 PM

  Emmanuel, I don't see where you would use this? I'm sure there is a reason, I just can't seem to figure it out. Regardless, the patch applies cleanly, but a) due to my thickness, I don't know what tests to write for this, and b) I'm not sure we should include the credit to Logan. It'll just get taken out one day, and since he posted it to a mailing list I'm sure he won't mind us using it.

  But other than that, it definitely didn't break anything!

• You flagged this item as spam.
  

  Emmanuel Surleau (Mercurial Alchemist) February 11th, 2008 @ 02:24 AM

  Michael,

  Thanks for your quick answer. I have uploaded some simple code which shows the problem. Basically, without the patch, you cannot use something like:

  :has_one :accessor_name, :class => 'ModuleName::ClassName'

  I'll be sure to include an rspec test next time, just don't quite have time to write it this morning :)

  • test_namespace_module.rb 1.1 KB

• You flagged this item as spam.
  

  Michael Boutros February 11th, 2008 @ 03:59 PM

  Yep, that definitely broke before, but your patch fixes it up. I refactored it a little bit and added a test. The patch is also for git, so if you're using SVN you might have to manually do the edits.

  • nested_modules_support.diff 3.5 KB

• You flagged this item as spam.
  

  Adam French February 11th, 2008 @ 06:46 PM

  • State changed from “new” to “open”

  emmanuel,

  does that :has_one association work when you pass :class as a Class, not a string? I'm thinking that strings get sent through the inflector and things fail because it attempt to singularize it to ModuleName::ClassNam

• You flagged this item as spam.
  

  Emmanuel Surleau (Mercurial Alchemist) February 12th, 2008 @ 01:14 AM

  Michael, thanks for incorporating this change.

  Adam, I didn't even know you pass a class directly. Unfortunately, the

  error stack goes like this:

          from ./../data_mapper/lib/data_mapper/dependency_queue.rb:15:in `resolve!'
          from ./../data_mapper/lib/data_mapper/dependency_queue.rb:14:in `each_pair'
          from ./../data_mapper/lib/data_mapper/dependency_queue.rb:14:in `resolve!'
          from ./../data_mapper/lib/data_mapper/dependency_queue.rb:10:in `add'
  

  If you use a Class directly, the method 'associated_constant_name' will

  take it and return the name of the class. It's the result of this method which gets passed to 'add'. By the way, looking at the code (unless I'm missing something), it seems a bit silly to do Class -> String (in HasNAssociation#associated_constant_name) -> Class (in DependencyQueue#resolve!).

  While looking at HasNAssociation#associated_constant_name, I noticed another, unrelated issue. The :class and :class_name parameters are equivalent, which means that :class_name => MyClassObject works. I'd suggest that DM should assume that :class is a Class object, and :class_name is a String object. Maybe add a type check when retrieving :class, and print a meaningful error message if it's a String.

  Just my two cents.

• You flagged this item as spam.
  

  Sam Smoot March 17th, 2008 @ 10:42 AM

  So wait, is this applied or not? Should it be closed out?

• 

  Dan Kubb (dkubb) April 15th, 2008 @ 03:36 AM

  • Milestone cleared.

• You flagged this item as spam.
  

  Sam Smoot April 27th, 2008 @ 10:59 PM

  • State changed from “open” to “resolved”

  Ok, I'm going to assume this is resolved since that's what it sounds like. If not, feel free to reopen the ticket.