On Wed, 6 Apr 2005 08:15:23 -0400, Ben Hinkle wrote:

>> I've got it down to only 5 (five) times slower now ;-) Mainly through some
>> optimization (removing redundant copying). I'm now playing with the
>> 'package' attribute to enable further optimizations, but at the cost of
>> increased module bindings (i.e. increased maintenance costs).
> 
> I'd be surprised if declaring functions as 'package' significantly improves things. Have you tried the -gt flag to see where the bottle-neck is?

Duh? I know that *just* by using the 'package' attribute one doesn't magically get any performance. But by using it, I can have one class directly see the contents of another without having to be polite about it. So instead of class Foo asking for a copy of Bar's data, Foo can just go in and grab what it needs.

And yes, I've used the -gt to get the improvements I've so far made.

-- 
Derek Parnell
Melbourne, Australia
7/04/2005 1:54:04 AM

Georg Wrede says...
>
>Ant wrote:
>> So, how many of you use the OO paradigma?
>
>I use it with practically all programs.

Ditto...

>But it's not a religion. It's just practical for some things, and impractical for others, where I don't use it.

Ditto...

>Mostly it just "comes by itself" whether or not.

Ditto...  It needs to feel right.

Sometimes people get into a OO Religion where they would make a class for something that would make no sense like i = 3;  And they write 5 lines of code to write something that could be done in one line of code.  So, I love OO, but when it applicable, "but it's not a religion," as Georg said.  Back when I started coding, 1981, we use to change codes of the apple IIe machine language games to make them faster or slower, etc.  I did some c, back then and PL/I and Pascal and RPG.  None of these were OO languages.  But, I like to welcome positive changes and OO is a positive change for programming.  But not in a way that **everything** needs to be OO.

Am I talking too much? :-)  Sorry.

jicman