[Ktechlab-devel] test-cases using KDevelop::Core

[Julian Bäume]

On Thursday 15 April 2010 21:33:45 P Zoltan wrote:
>   Then getting the model is currently not possible. Not a big problem, can
> be fixed with the interfaces.
Yes, I will add it to the IDocument interface today. Patches will follow, 
soon.

>   Now I'm thinking how hook up the simulator.
> 
>   The debugController might be useful for running PIC programs. It might be
> usef for circuits, too, but I don't know which one is the better solution:
> implement the IDebugSesstion interface or subclass the Core and add
> SimulatorController type member, where the documents can register
> themselves. It's not obvious.
I thought about implementing some kdevelop interface, too. IDebugSessions 
sound promising, also the ILaunch* interfaces could be interesting.
 
>   For simulation we need:
> - the actual simulator, for different document types
I'd suggest different simulator plugins that are loaded for different mime types 
(of the document). So the simulator does nothing like: if (document.type() == 
"foo") { //bar } but there is an abstract factory and you ask it to create a 
simulator for a document with mime-type circuit.

> - a way to register new component models for a simulator class/document
> type, in order to be loaded from plugins
> - the models for the compoennts
New models can also be provided by plugins. We should have a basic 
ComponentModelPlugin interace and the simulator just uses the PluginController 
to retreive a List of plugins that provide a model for component X.

>   So one way to do this might be:
> create an ISimulatorController. It contains a list of registered component
> types and the factories associated with them, and also a list of document
> types and document simulator factories.
> When a new document is created, it creates a new document simulator with
> the document type (great, now we will again have document.type()), and
> attaches the simulator to the document model. Then the document simulator
> on each model change can rebuild the structure formed by the models. Also,
> when a given time elapses, the document simulator runs a step in
> simulation.
I don't think a SimulatorController will be necessary. At least no new 
interface might be needed for that. I'm not sure, but we might be able to 
implement some existing interface to act as a SimulatorController. I haven't 
thought much about that part, yet.

bye then
julian