Re: [GDL2] commit and few points

[David Ayers]

Manuel Guesdon wrote:
> But doc: 
> http://developer.apple.com/documentation/LegacyTechnologies/WebObjects/WebObjects_4.5/System/Library/Frameworks/EOAccess.framework/ObjC_classic/Classes/EOAttribute.html#//apple_ref/occ/instm/EOAttribute/adaptorValueByConvertingAttributeValue:
>
>
> says that - (id)adaptorValueByConvertingAttributeValue:(id)value 
> Ensures that value is either an NSString, NSNumber, NSData, or
> NSDate, ...
>
> As far as I uderstand, it doesn't say that value should be of the
> same type than declareted in EOAttribute.
>
> But I may be wrong. How do you undestand it ?

Indeed the wording is not great but what is meant is that it will 
convert the value to the /corresponding/ class defined by the attribute 
type.

If you try to pass an NSNumber to an EOAttribute set to 
EOAdaptorCharacterType in WO45, you will get an exception saying:
"... unable to convert value of class NSshortNumber for attribute in 
entity Book to adaptor type EOAdaptorCharactersType"

And this is what I would expect.  It's there to avoid the the following 
scenario.  Take a custom class that declares an NSString ivar.  Then 
have KVC set the attribute to an NSNumber.  Then call some method that 
uses the ivar and is expecting an NSString.  You should expect undefined 
behavior.  In fact if you attempt to sort the EO's according to this 
attribute with compare:, I'm almost certain you get a crash, as compare: 
is only safe for compatible objects (the implementations often access 
ivar's directly).

Cheers,
David