[kde-edu]: Step and Summer of Code ?

Vladimir Kuznetsov ks.vladimir at gmail.com
Thu Mar 15 19:47:18 CET 2007


Hello Ralf !

On 3/15/07, RalfGesellensetter <rgx at gmx.de> wrote:
> Dear Vladimir,
>
> only a brief reply:
>
> Am Donnerstag 15 März 2007 16:44 schrieb Vladimir Kuznetsov:
> > 3d is quite more complex both for GUI and for simulation (I mean
> > rigid bodies and collisions), so I'm not going to start working on it
> > until 2d will be actually usable.
>
> This is true, of course, and many models work alike in a 2D reduction
> (planets spinning around a sun, for instance).
>
> This doesn't mean that it couldn't be wise to prepare for dimensional
> extensions by using general classes/coordinates rather than hard-coding
> 2D. I try to compare this to localisation: it is harder to prepare
> written code for other locales afterwards than being prepared from the
> very beginning.
Yes, I'm trying to make StepCore as dimensional-independent as
possible. Currently basic classes are still not generic, but mainly
because I still doubt about how to make API consistent, technically
they can be made generic quite easily. Some classes in StepCore does
not depends on dimension at all (World and all Solvers).

But there is also large part which is quite different (and quite more
complex) in 3d than in 2d: rigid bodies and collisions. Angle, angular
velocity, moment of inertia are all scalars in 2d, but in 3d they
should be represented as tensors (however this is not the only one
possible representation...). Collision detection and handling is quite
more complex too. All of this also implies that in 3d there are lots
of effects that simply does not exists in 2d, for example gyroscope
(btw in quantum mechanics 2d and 3d cases are even more different).

> When I talk about multidimensional models (do you know k3dsurf BTW?),
I've looked at k3dsurf: it looks great ! Why this program is not in kdeedu ?

> I
> don't mean 3D graphics (this is yet another issue): Noneregarding
> internal models dim., all states should be projectable to 2D graphics,
> I think.
In my opinion, simply showing one projection without ability to rotate
will give user wrong impression about experiment: particles that is
far away from each other can look closer, moving particle can look as
non-moving, etc.

First step to 3d can be "read-only" simulation that user can view (and
can modify only fixed amount of numerical parameters) but can not
create using GUI. It still not requires implementing 3d editor, but
already can show lots of cool things (for example 3d molecules, this
is what I mentioned in previous post). Then we could implement 3d
rigid bodies, and then 3d editor. But I think that first of all we
should make usable 2d version.

-- 
       Best Regards,
          Vladimir


More information about the kde-edu mailing list