[Ktechlab-devel] A few more ideas

[P Zoltan]

On Mon, 14 Jul 2008 08:24:38 +0200, Alan Grimes <agrimes at speakeasy.net>  
wrote:

>
> I had taken a slightly different approach. Since certain things must
> always be true for a circuit (eg, all currents must sum to 0, etc..),
> the voltage on both sides of a wire must always be equal, it is possible
> to implement runtime checks. Several of my recent versions will print
> out some of those errors. The trick is to design the code such that it
> will be difficult for these bugs to appear.
>

  Kirchoff's first and second law and the component's equations should be  
enuogh to verify if a solution is valid or not. The problem is the CPU  
usage generated by these checks -- in a debug version I would use it, but  
not in a release/"production" version

  Another problem is that when playing with the visualisation part, there  
might happen that the simulator "sees" different circuit than what  
appeares on the screen (considering that mess with the ECNode, Connector,  
Item, ...). Dumping the circuit as a netlist would be nice.


> yeah...
> I like the idea of moving to overloaded type() first, that will help to
> clean up the class heirarchy... Then we can fix the other stuff.
>

  I've downloaded the revision 103 from svn and I'm starting to code the  
proposed stuff. I'll send some patches when it's done.

  A note here: to compile it successfully, I had to change in the file  
src/math/Makefile.am , the lines:

libmath_la_SOURCES = qvector.cpp qmatrix.cpp

libmath_la_LIBADD =

  to:

libtechmath_la_SOURCES = qvector.cpp qmatrix.cpp

libtechmath_la_LIBADD =