-hash is broken on CPArray

Why do both array's need to return the same hash value? That doesn't seem to be documented anywhere in the Cocoa docs.

This code seems incredibly dangerous to me. You're essentially saying that all kinds of arrays that have nothing to do with each other could be considered equal. This will break anything that uses hash as a unique identifier (for example, KVO/Bindings, and I'm sure other code in Cappuccino).

I spoke before I understood what I was talking about.

From the documentation for hash:

@"If two objects are equal (as determined by the isEqual: method), they must have the same hash value. This last point is particularly important if you define hash in a subclass and intend to put instances of that subclass into a collection."

Our implementation of hash does not behave this way. Perhaps we should consider changing our hash to become address, and re-implementing hash the Cocoa way?

I'm not sure I understand you correctly, -[CPObject hash] does return __adress, I haven't checked all the other objects, but all objects who redefine -isEqual: need their own implementation that matches the contract.

The one which I provided for CPArray - and I checked it against the two other Cocoa implementations I know, which do it the same way. (Also it makes sense: The Implementation needs to be FAST, and it needs to be the same when two objects are -isEqual:

I also checked against some examples on Cocoa and it behaves the same way.

My point was that -hash works differently in Cappuccino than in Cocoa (something I previously had not realized).

We need access to __address as a method, which is what hash has been doing for us. To make -hash actually correct, we'll need to add -address and update all of the cappuccino code that relies on hashes being globally unique.

Too late to finish sentences...

The one which I provided does fit the contract and also behaves the same way that the others do.

Yes, but it breaks parts of Cappuccino that expect the current behavior. For example, KeyValueObserving and CPWindow both rely on the value of -hash as a unique index into various data structures.

Oh damn.

I just had a look around in the source and lots of objects like CPDictionary, CPData, CPDate and CPNull don't overide -hash but should. (And most should probably also overide isEqual:)

Uh oh....

Yes, this is going to take some work.

Can't resist...

As a good person once said: In God we trust - for everything else we have UnitTests.

:)