Levente Uzonyi wrote:
>
> Okay, just for nitpicking, the #waitForSeconds send is unnecessary in 
> DateAndTime >> #now, because #milliSecondsSinceMidnight will send it.

Right. I left it because I felt it would be good to state that the 
offsets must be consistent... I guess a comment is enough. I'll remove it.

>> Right. I did not remove #detect:ifNone: as that would break a lot of 
>> code. But I believe having the actionBlock in the sender will usually 
>> save some #ifTrue: to do stuff only when an element was found.
>
> The only minor problem with blocks is that they have to be copied and 
> activated which is sometimes too expensive.

Right. I guess anybody writing such performance critical code for this 
to matter, should check all the called methods himself, and tweak as needed.

> But, your exceptionBlock enhancement is fantastic in 
> #findBinary*ifNone:. I was about to implement some kind of C++-style 
> #upperBound: and #lowerBound:, but your solution already includes that 
> and it's even more general. It shows the power of Smalltalk and blocks.

:) I also liked that one!

>> Cuis change sets are numbered and should considered in order. Latest 
>> update number is 034, so 431 is quite old. It is actually older than 
>> Cog public release. That code was modified later, when the issues you 
>> mention were found in Squeak. You can check that by opening a changes 
>> list on change set 431 in Cuis. MessageTally works correctly in Cuis 
>> on Cog.
>
> I'm aware of the numbering scheme, but I was checking the changesets 
> by name. The fixes are in 0511-CogCompatibility and I was only looking 
> at the changesets that had tally in their name.

I see.

>
> Levente

Cheers,
Juan Vuletich