[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-kimageshop
Subject:    Re: For the next release
From:       Boudewijn Rempt <boud () valdyas ! org>
Date:       2005-08-11 14:31:45
Message-ID: 200508111631.45369.boud () valdyas ! org
[Download RAW message or body]

[Attachment #2 (multipart/signed)]


On Thursday 11 August 2005 16:28, Bart Coppens wrote:
> On Thursday 11 August 2005 16:16, Boudewijn Rempt wrote:
> > If you use a timer, and every time the timer fires dry only a 10x10 area
> > of pixels, and make sure the areas you dry are distributed you might be
> > able to fake multithreading.
>
> I don't think the physics code can be easily distributed among tiles. From
> what I remember, the drying itself is a simple 'water-=1', but the entire
> processing that comes before that is a lot more complicated than that.

Even when running in a thread it may be important to distribute the activity 
over the canvas, instead of doing it scanline by scanline. Cockshott warns 
explicitly against the windscreen-wiper effect.

> That sounds very bad. So multithreaded krita will definitely need a lot
> more investigation to do. And that's probably not the best thing to spend
> our time on if we want to do a release in the near future.

Yes -- it's a bit harder with C++ than with Java, I discovered. But Michael 
and I spent only one evening on it, not more -- and there may be an easy way 
to serialize access to areas of interest and enable reading from the olddata 
from one thread while writing to newdata of a single from another thread.

It's interesting stuff, but as you, say, if we start playing with that, we'll 
never release, or at least not soon.

-- 
Boudewijn Rempt 
http://www.valdyas.org/fading/index.cgi

[Attachment #5 (application/pgp-signature)]

_______________________________________________
kimageshop mailing list
kimageshop@kde.org
https://mail.kde.org/mailman/listinfo/kimageshop


[prev in list] [next in list] [prev in thread] [next in thread]