On 13 March 2010 12:20, Stephan Eggermont <[email protected]> wrote:
> Levente wrote:
>>How bad do they perform? Do you have benchmarks?
> With a
> coll := OrderedCollection new.
> 1 to: 10000000 do: [:i | coll add: i].
>
> a profile of:
> 1 to: 100 do: [:i | coll add: i beforeIndex: 10]
> here takes 1438 ms. Moving memory gets to be slow.
> All operations that start taking O(#elements) time are
> no longer funny.
>
> The usual thing to do is to start using multiple blocks
> of memory (like a BTree). Doubling size when growing
> is also not an acceptable strategy when getting close to
> total ram capacity.
>
Yes. Use right tool for to do job.
An OrderedCollection can't satisfy every possible combination of
tasks, which developer facing.

>From this point, i barely see how we can achieve a substantial
improvements in current implementation.
And i think if we talk about collections, we should be focused more on
a refactoring (use traits) for better
library design, which will make implemetation more compact and cleaner.

> Stephan
>
> _______________________________________________
> Pharo-project mailing list
> [email protected]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>



-- 
Best regards,
Igor Stasenko AKA sig.

_______________________________________________
Pharo-project mailing list
[email protected]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Reply via email to