Traits approaching mainstream Squeak

Marcus Denker denker at iam.unibe.ch
Wed Aug 31 19:08:09 UTC 2005


Am 31.08.2005 um 20:41 schrieb Herbert König:

> Hello David,
>
> DPH> Why do you say this?  There are ways and means to deal with  
> this - if it
> DPH> is actually a significant problem.  Time saved in
> DPH> development/maintenance may mitigate this, also.
>
> Software being said to be sluggish is in a big distvantage. In road
> design where I plan to use squeak commercially there is a speed
> difference between 3.6 and 3.7 and not using accessors and pushing
> methods down the hierarchy makes a difference.
>

I don't think that the speed difference 3.6 vs. 3.7 due to using  
accessors.

I am sure that the morphic cleanup changes (isEmpty vs. direct  
testing and so on)
had some influence, but the majority of the speed-loss is due tue the  
different
window look, esp. the buttons with gradients, using true-type for  
window titles
and stuff like that.

In 3.9, performance is now better than 3.7, but not yet as fast as 3.6.

Overall, I have the feeling that real improvments in performance will  
not come
from micro-optimizing, but from using better algorithms and better  
concepts.
And the first step of this is cleaning up.

> And though I like LookEnhancements I always notice the speed penalty.
>

On my system, 3.9 with LookEnhancements is *faster* than the 3.9 without
(3.9a6686 vs. 3.9a6684). Can you make a test? Would be interesting to  
see.

     Marcus


More information about the Squeak-dev mailing list