> Probably not a great idea, but certainly no catastrophe either.
> Numerical precision problems are best solved by understanding and
> mitigating them, not clubbing them over the head with extra
> precision. :)

Very, very true. My concern is that we are using about half floats and half 
doubles. I am thinking that there are some cases where very small numbers that 
are converted between the two are causing problems. Maybe not. If there are no 
*performance* penalties with moving to all doubles I'd be interested in trying 
that - *just* to see what happens. Probably nothing will, I'm just curious. 
But, IIRC the standard library math functions all return doubles in 
their "normal" forms, i.e. cos, sin, log, pow, etc. all return doubles. I 
thought it would be better to use "natural" data types. No?

Jon




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

Reply via email to