Re: [Jsbsim-devel] P51D

[Dennis J. L. <js...@li...>]

On 8/14/10 5:46 AM, Jon S. Berndt wrote:
>> I would suggest that there shouldn't be a default value, rather an
>> warning if the attribute is zero length, assuming this indicates either
>> an empty, misspelled, or missing "logic" attribute.
>>
>> If they had actually meant "OR" in the case above, it would have taken
>> a
>> *long* time to find.  I think silently "accepting" misspelled keywords
>> is not a good way to go.
>>
>> Looking at the manual, the logic attribute in the <test> statement of a
>> <switch> is not indicated as optional, nor does it mention a default.
>>
>> The logic attribute of a <condition> statement is indicated as optional
>> with a default value of "AND".  I don't think this is a good idea for
>> the same reason, but it is documented.
>>
>> Any comments?  Should I patch the above code to match the manual?
>> Clearly the p51d files need to be fixed as well...
>>
>> Dennis
> 
> Good points. It is of course obvious that we should validate the input files
> before commencing a run. The question is how do we do that, and where does
> the responsibility belong for any validation? In my opinion, that's where
> the XML Schema comes in handy. The current schemas need to be updated. And
> also a command line validator is needed, such as xmlstarlet. It would be my
> preference to validate JSBSim input files using an external command line
> tool just prior to running JSBSim, than to litter JSBSim code with tons of
> validation code. It's hard to tell for sure what the balance should be. 
> 
> Suggestions?
> 
> Jon

You make even better points Jon.  It would probably be best to validate
appropriately to a (updated?) schema instead of adding such code to
JSBSim.  That's one of the main points of XML after all.  I got a little
too caught up with "fix the warnings and errors" as I sometimes do,
without thinking about the bigger picture.

Dennis