[sylpheed:36297] Re: A performance problem (rant).

[Onur Küçük]

On Sun, 22 Mar 2015 10:41:13 +0200
John Found <johnfound at asm32.info> wrote:

> On Sat, 21 Mar 2015 19:51:33 +0200
> John Found <johnfound at asm32.info> wrote:
> 
> > I got it. It works. Well, it is some kind of cheating for me, but 
> > better than nothing. ;)
> > 
> 
> BTW, it is a cheating, because so big performance degradation shows
> some sub-optimal computational complexity (big O(?)) that probably
> need to be fixed.
> 
> One more observation: if the user hides the search panel, the speed
> degrades instantly.
> 
> IMHO, the speed of browsing need not to be tied to the searching
> features. These two features of the client are different and need
> different approach. For example, searching something (by the "search"
> feature), the user usually accepts some delay, but delay during
> manual browsing (which is actually a visual search by the user
> itself) affects the user performance and this way is not acceptable.

 From the start of this thread you seem to be stuck at "computational
performance", comparing CPU performance etc. But I believe you should
start with hard disk and filesystem performance of your system, and
then move to how your OS handles filesystem and  memory (for
caching filesystem). If I were  looking into how I can make this faster,
I would do some profiling on both CPU and IO, and check all kinds of
resources to figure out where things get slow / where the bottleneck is.

 I am also quite happy with the performans of Sylpheed and for all
these many years I don't remember ever using the options "last 7 days"
or similar.

-- 
 Onur Küçük                                      Knowledge speaks,
 <onur.--.-.delipenguen.net>                     but wisdom listens