[Ktechlab-devel] unit test.

[Alan Grimes]

P Zoltan wrote:
>   Finally I have some time for this issue. Looking through the source, it  
> seems a litte under-documented (inline) for me and also I'm not very happy  
> with the name of different identifiers; first I'd like to add some  
> comments and doxygen documentation. What kind of coding standards should  
> we use? Something like this would be good:  
> http://techbase.kde.org/Policies/Kdelibs_Coding_Style .

Sure. =)

>   About tests: I don't want to create a testing framework here, but I would  
> be more useful to create some test types (some class), holding the set of  
> operations and particularize them for various input data in test cases.  
> That should catch most of the bugs.

ok... I think I've nailed most of the easy bugs in the calculation
engine, the process seems to have exacerbated bugs of various kinds in
BJT, jfet, and especially mosfet. =(


>   A final question: should limtechmath (and all the test cases) depend on  
> kde libs, or not? It can be easly done, because only the way the output is  
> printed must be changed. The only problem is that, as I've already  
> mentioned it, that cout mixed with kdDebug() can create inconsistent  
> output. Any ideas are appreciated.

"deep" code should always output to a "FILE" or stream object, that way
it can be redirected to whatever purpose is desired by the front end.


-- 
New president: Here we go again...
Chemistry.com: A total rip-off.
Powers are not rights.