Laurent Bossavit wrote:
> Marshall,
>
>>My general claim is that one line of code is typically
>>*not*a coherent abstraction. Because once we abstract it
>>as a method, what do we need to do to invoke that method?
>>One line of code.
>
> I would suggest you count the number of message sends rather than the
> number of lines.
>
> So, suppose we replace a line of code containing a complex expression
> (say, two additions and one multiplication, for a total of three message
> sends) with a single call - we can claim to have raised the level of
> abstraction.
You reveal an inability to think in the abstract. A declarative language
is fully optimizable right down to the physical hardware. Two additions
and a multiplication need not 'send' anything. In fact, depending on the
context and what one is adding and multiplying, the compiler might not
generate any executable code for them at all.
Received on Tue Jul 04 2006 - 20:50:02 CDT
