On Sat, 2004-01-31 at 04:46, Mathias Fr�hlich wrote:
> On Samstag, 31. Januar 2004 12:46, Christian Mayer wrote:
> > Jon Berndt schrieb:
> > > I just discovered that our property "metrics/eyepoint-x-ft" shows that it
> > > is expected to be in units of feet. However, it is specified in inches in
> > > our config file (in structural frame), and the property is bound to the
> > > GetXYZep() function, which reports the eyepoint in the same way it is
> > > specified: units of inches, in structural frame. I suppose this ought to
> > > be changed.
> >
> > If we'd use SI units everywhere we wouldn't have that prolem...
> > (and if NASA would have consitently used SI they wouldn't have lost one
> > Mars probe...)
> I am also used to the SI system.
> But all I have seen up to now in the area of physical aircraft models is given 
> in units of feet and such. So I think the FDM should be able to work using 
> this values.
> 
> What I can think of is an extension to the property tree, either private to 
> JSBSim or better in Flightgear in general.
> I would like to be able to tie a value with a given unit into the property 
> tree. Then there should automaticaly appear a set of values in the property 
> tree which represent this value in different units.
> 
> There are many possible scenarios how this can happen. The most compatible one 
> will be to tie a value with unit into the property tree without the unit 
> extension in the name. Then there are automatically the usual unit values in 
> the tree. For example the angle of attack will be tied to the variable 
> 'alpha' with an angular unit originaly given in rad, but that alpha will not 
> appear as 'alpha' in the tree, but instead two variables appear. One 
> 'alpha-rad' and one 'alpha-deg'. The same can of course be done for unit 
> conversion from and to the SI system.
> The other possibility, which is not that backward compatible, will be that 
> each value is represented by a whole directory. Inside this directory there 
> are then only unit names. For example the angle of attack will then be 
> accessible through alpha/deg or alpha/rad.

I certainly do agree that selectable units would be nice to have, but it
would double the property tree memory requirements (since each property
would then have to have a units property associated with it. In
addition, such a system would significantly increase the amount of code
that would execute at each property access.

> 
> This will also simplify building aerodynamic coeficient lookup talbes which 
> are sometimes given in, say 'lift per degree elevator deflection' or, for an 
> other aircraft type in 'lift per rad elevator deflection'.
> I know, this is simple scaling, but I think that simple things should be left 
> to a computer ...
> :)

All coefficients must be in consistent units (generally those of force)
after the multipliers are applied, so I see no advantage here.  There
is no formal requirement for this since JSBSim does not attempt to
assure unit consistency but it's there nevertheless.
 
> 
> If someone has spare time to work on such a automatic unit system ...
> Also, I have not looked into the property tree implementation if this could be 
> done easily.
> 
>    Greetings
> 
>      Mathias
-- 
Tony Peden <[EMAIL PROTECTED]>


_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Reply via email to