Hi Keith,

comparing the selector of two block's home is, IHMO, like comparing a  
person's name with its nickname. Also a CompiledMethod might not have a  
meaningful selector (for example if it's as yet not in a MethodDictionary,  
which is the case since Compiler's DoIt evaluation was hacked [recently]).

I'd suggest a more general form, replacing that line by

  self home method = other home method ifFalse: [^ false].

which ensures equal blocks would execute the same bytecode.

Moreover, the two blocks could have been created from different  
MethodContexts, so an even more general form (which implicitly compares  
the receiver, args + temps and the method) would be

  self home = other home ifFalse: [^ false].

/Klaus

On Mon, 24 Dec 2007 02:29:59 +0100, Keith Hodges wrote:

> Tom Phoenix wrote:
>> On Dec 23, 2007 3:11 PM, Keith Hodges <keith_hodges at yahoo.co.uk> wrote:
>>
>>
>>> I am wanting to test the equality of blocks, as are the two blocks that
>>> I have defined in the same method context.
>>>
>>
>> Are you talking about using #== here? Two blocks are equal if and only
>> if they are the same block? Or is there some other kind of block
>> equality you're looking for?
>>
>> --Tom Phoenix
>>
> According to  http://bugs.squeak.org/view.php?id=6599 you can test
> BlockClosures for equality.
> However I had to define my own test for BlockContext's
>
> This does what I want
>
> BlockContext>>#= other
>
>     self class == other class ifFalse: [^ false].
>     self home receiver == other home receiver ifFalse: [^ false].
>     self home selector == other home selector ifFalse: [^ false].
>     ^ self startpc == other startpc
>
> TestBlock>>#blocks
>
> 	^ Array with: [1] with: [1]
>
> TestBlock>>testBlocks
>
> 	self assert: (self blocks = self blocks).
> 	self deny: (self blocks first = self blocks second).
> 	self deny: (self blocks first == self blocks first). "**"
>
> ** a #== comparison will not succeed since although it is actually the  
> same block, the instance of blockContext is different
>
> Keith
>
>
>
>
>
>
>