[Ktechlab-devel] simulator

[Celelibi]

2008/11/9 Glen <gcanaday at gmail.com>

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
>
> There are still a lot of things missing in KDE 4.1. There appears to be
> no way to put a random app from the menus into a panel; apparently they
> all have to be plasmoids, and you cant use a random plasmoid. The
> wireless / network monitor cannot be moved to a different panel... in
> fact, more than half the default icons cant be moved. I cant change my
> keyboard layout. Im stuck with deadkeys I cant use... hence no quotes in
> this message and no apostrophes.
>
> Positive: intrepid has kernel 2.6.27. My wireless works beautifully now.
>

<troll>
So, you use ubuntu...
Ubuntu sux, it's too much user friendly. :p
</troll>


> Serious negative: KTechLab will not compile. My bad, I likely dont have
> the KDE3 dev libs installed anymore.
>
> Getting back to the discussion though, I want to see the simulator have
> its computation built in (even so far as replacing a subcircuit with an
> expression if not on the same sheet if desired to speed up computation),
> so it will need an equation editor. These equations could be saved in
> the subcircuit file. The XML is strictly a simple way of organizing the
> file format IMO, so that the component attributes, the SVG, the
> footprint, and if need be a (simple!) 3D model of the component can be
> included in the library.


I'm not sure to understand what you want to see inside the core of de
simulator.
The core of the simulator must be quite small, and equation solving should
be placed in some other module/class.
If you're ok with that, I just ask : why don't buid these modules as shared
objects ? This would simplify addition of new solvers.

About embedding equations inside the XML. It will be possible describe every
component with equations only if we can define our own functions.
But defining a function is useless if you don't know how to solve equation
containing it. So you need to define a way to solve any equation containing
that function. But maths says it's not possible.
Let me explain with a small example.
You have a dipole for which current/voltage is such as u=f(i)+g(i). Of
course, you want to calculate i from u. If f is a polynomial and g is
something that deals with logarithm or exponential, a generic solver will
have some pain to find something.
What I try to explain is that embeding equation inside a component
definition will lead to unsolvable problems.

I think a more powerful solution would be to make the XML say something like
:
MyComponent behave following That_Model with parameters "voltage in pin1",
"current in pin2" etc.

I really really think embedding equations is a bad idea.


Celelibi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/ktechlab-devel/attachments/20081110/8dd63976/attachment.html>