Re: ADC Core Data article

FROM : mmalcolm crawford
DATE : Wed Apr 06 23:47:06 2005

On Apr 6, 2005, at 1:35 PM, Nat! wrote:

>> WebObjects 5 certainly requires that you inherit from 
>> EOCustomObject; EOF 3 (in WebObject 4) in effect required that you 
>> inherit from EOGenericRecord:
>>

> Well on the Java side maybe, but not in Objective-C.
>

Indeed -- mea culpa.
The EOEnterpriseObject protocol was implemented as a category of 
NSObject.  That said, I tended to subclass EOGenericRecord anyway 
since it made the entity more flexible during early stages of 
development...

The big picture issue remains, however, that Core Data is not the 
same as EOF.  It's clearly addressing a different problem domain 
(primarily desktop applications with local files, as opposed to 
enterprise-level client-server applications which is what EOF evolved 
to address) and has a different set of constraints.  Comparison with 
EOF may be useful for historical interest -- and (primarily for those 
curious at the moment but who don't have access to Tiger) to get an 
idea of the likely architecture -- but thereafter it ends.  I think 
it's reasonable to assume that the engineers thought long and hard 
about what problems they needed to solve and what trade-offs they had 
to make.  And about what they had learned from the past.  What's 
being delivered in Core Data is not, as some would appear to want to 
suggest, a stripped down version of a previous product, but instead 
an evolution that meets a different need...

If you happen to have used "earlier generations" and believe that 
Core Data would benefit from features that are not present in the 
release, then please file enhancement requests.

mmalc