----- Forwarded message from Youness Alaoui <[EMAIL PROTECTED]> -----

From: Youness Alaoui <[EMAIL PROTECTED]>
To: Nathan Ingersoll <[EMAIL PROTECTED]>
Subject: Re: [E-devel] Starting programming in EFL

Exactly, I don't care much about the performance (as long as it's not affecting 
usability). I was just answering Michael that his "EWL has better 
performance, period." is a bad argument since it is unproven facts. And I don't 
care about those 'facts' since ETK devels could say the same 
thing, and I was pointing out that the EWL_Performance page was not very 
significant if we don't plan to use thousands of widgets.


On Mon, Sep 10, 2007 at 09:39:49AM -0500, Nathan Ingersoll wrote:
> On 9/9/07, Youness Alaoui <[EMAIL PROTECTED]> wrote:
> >
> > I'm not talking here about what the human eye/brain can perceive. But if 
> > you want more details. I was refering to the wiki page about EWL
> > performance. That's one of the main arguments used by the EWL team, and it 
> > is true that EWL looks MUCH more performant if we reason by looking at
> > those numbers.
> 
> It is? I don't recall ever telling someone that performance is the
> main reason to use EWL. I'm fairly certain I told you to take those
> numbers as a way we are testing widget scalabiity. The other toolkits
> are there for a frame of reference, since it's hard to know what is
> "good" or "bad" without a frame of reference.
> 

Maybe not you, but I've heard others.. 

> > But seriously, how many applications use 100K widgets ? What are the "real" 
> > performance numbers ? If we compare each and every
> > widget provided by EWL and ETK, which ones are faster?
> > what if we use 10 widgets and not 100K widgets? which toolkit is more 
> > performant ?
> 
> "Faster" doing what? Setup, layout, repainting, scrolling, etc, etc?
> What theme are you using? How many different images or fonts are used?
> What features are you using for those particular widgets? There are
> many things you can test for performance. I took the time to test and
> optimize one case, but I have found that test useful for improving
> other areas as well. I barely have time to work on EWL code let alone
> test and compare every single aspect of various toolkits.
> 

Yep, that's exactly what I had in mind, being able to see the time (and memory) 
used for each widget, for each operation, creating it, 
configuring, repaiting, etc.. I might do it myself, I'm not asking you to do 
it, I'm just saying that unless documents like that are provided, I 
don't think that saying "X is more performant than Y" is a valid statement.

> > That test was merely a scalability test, that's all it is. I don't plan on 
> > use ETK or EWL in such proportions. If I needed to create a chart
> > with hundreds or thousands of buttons and labels then ok, EWL is better, 
> > but for a "normal" application, I don't know which one is more
> > performant that the other, and even if EWL is still more performant, if 
> > it's a 5% increase in performance, is it worth it to choose EWL over ETK
> > because of performance, if overall, EWL responds less to our demands...
> > That's what I'm talking about, so please, no need to use such an argument.
> 
> Why are you so concerned about performance for such a small case?
> Unless the toolkit is causing blocking points, busy loops, or doing
> something else stupid, there will most likely be very little
> difference.
> 
> > And by the way, quicksort might be faster than bubble sort, but only for 
> > large numbers of 'n'. I'm sure that bubble sort is faster on smaller
> > samples. So if you want to use a sorting algorithm, you would see "if 
> > you'll work on small samples, use bubble sort, if you plan on sorting huge
> > samples, then use quicksort", so if you know in advance that you'll never 
> > sort a list of more than 5 elements, why use quicksort ?
> 
> And unless you are repeatedly sorting this small data set, how is it
> of any interest? The difference is going to be barely measurable on a
> modern system.

Exactly, I don't care much about the performance (as long as it's not affecting 
usability). I was just answering Michael that his "EWL has better
performance, period." is a bad argument since it is unproven facts. And I don't 
care about those 'facts' since ETK devels could say the same 
thing, and I was pointing out that the EWL_Performance page was not very 
significant if we don't plan to use thousands of widgets.

Thanks for answering!
KaKaRoTo

----- End forwarded message -----

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
enlightenment-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel

Reply via email to