Magnus Lundin wrote:
> Dick Hollenbeck wrote:
>   
>> Magnus Lundin wrote:
>>     
>>> The tap state engine will not change any time soon.
>>>
>>> Hopefully the statenumbers will stay.
>>>
>>> But there will probably be work on shortest path tables, path 
>>> transition tables with fixed mumber of tap transition etc. For this 
>>> type of work I think an array is a better and less error prone 
>>> representation than a big set of nested switches. It is graphic :) 
>>> and that is good. I can also play with and debug array repsentations 
>>> of state graphs in other tools like Octave. This very good way to 
>>> check the validity and catch errors. Actually, now when I think about 
>>> this, it should be possible to generate the arrays with some 
>>> Octave/Matlab code and calculate path lengths.
>>>
>>> So my wote goes for arrays.
>>>
>>> Regards
>>> Magnus
>>>   
>>>       
>> How should we count your vote, with or without array out of bounds 
>> checking?
>>
>> Dick
>>
>>
>>     
> I am perfectly ok with array out of bounds checking, it will catch some 
> silly errors and on PC platforms we have sufficient processor cycles for 
> that.  Bounds checking can always be turned off when the code runs 
> stable and performance is critical. 

I think you just described an assert()

Has this been discussed in the past?  I have not seen any in the code.

Dick


> The really hard errors are logical 
> mistakes within a state transition sequence that always keeps valid 
> states but it does the wrong thing. Here my vote is for the most clear 
> and understandable representation.
>
> Regards
> Magnus
>
>   

_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to