Re: [Rubycocoa-talk] Totally stuck with Core Data on Leopard

[Pierce T. W. I. <pi...@tw...>]

On Nov 11, 2007, at 8:28 AM, Jacob Wallstr=F6m wrote:

> Summary of post: I can't get ruby subclasses of NSMangedObject to work
> if they have their own attribute implementations with RubyCocoa on
> Leopard. I desperately need this to work, have anybody solved this
> problem?
> ---
>
> I'm having major problems with porting my application to Leopard. The
> problem is that my application relies on ruby subclasses of
> NSManagedObject which contains methods implementing the core data
> attributes. The default attribute implementation provided by Core Data
> is not enough since changes in an attribute can trigger changes in
> other attributes etc. Since I have a UI that uses bindings, a lot of
> the access to my entities is happening from Obj-C and I need these
> calls to go through the ruby methods of my entities.
>
> I got this working quite well with RC 0.12.0 on Tiger. By overriding
> setValue:forKey: and having it check for a ruby set-method before
> passing the call on to the setValue:forKey: method in NSManagedObject
> I could get all access from Obj-C to the attributes to pass trough the
> ruby methods which was what I wanted. I also used KVO-observing to get
> notified of changes in certain attributes.

I had problems with this as well on Tiger, but my solution was to use =20=

MOGenerator to make Obj-C base classes for all of my data objects with =20=

the accessors on that side, then override them from ruby if I needed =20
too.

That also had the advantage of being faster because:

def value
	willReadValueForKey("value")
	result=3D @value
	didReadValueForKey("value")
end

Had several bridge crossings to deal with, especially when used with =20
bindings, so that sped up anything that just needed a simple accessor.

You could make your own subclass of NSManagedObject and override the =20
method that handles "unhandled keys" because on Tiger at least, that's =20=

actually how NSManagedObject works. setValue: forKey: falls through to =20=

being "unhandled" and then NSManagedObject uses its default =20
implementation.

For Leopard, I'm not sure how much help I can be because my app broke =20=

with weird Core Data errors. So I upgraded to trunk thinking it might =20=

be issues with the last version I was using vs. release, but now it =20
crashes inside rubycocoa instead. So now every morning I do:

   svn update
   ruby install.rb setup
   sudo install.rb install

With my fingers crossed but I'm getting:

2007-11-14 10:24:19.723 Frictionless[10512:10b] LIBFFI : arg[5] -> =20
ffi_type 0xa04b4034
2007-11-14 10:24:19.723 Frictionless[10512:10b] OVMIX : Registered =20
Ruby method by selector =20
'outlineView:willDisplayCell:forTableColumn:item:' types =20
'v24@0:4@8@12@16@20'
/Library/Frameworks/RubyCocoa.framework/Resources/ruby/osx/objc/=20
oc_import.rb:266: [BUG] Bus Error
ruby 1.8.6 (2007-06-07) [universal-darwin9.0]

Abort trap

  My apps is open source from http://www.twinforces.com/frictionless/ =20=

though if you want to check it out.

Pierce