Andy Ross wrote:
>
> Arnt Karlsen wrote:
> > Bernie Bright wrote:
> > > FlightGear compiles fine with gcc 3.0, haven't tried 3.0.1 yet. Don't
> > ..for Red Hat 7.1 too???
>
> FWIW, I had no trouble compiling with the gcc "2.96" shipped with Red Hat 7.2,
> but building with gcc 3.0.1 (build from source) failed. It seems to have changed
> the floating point casting rules, or perhaps the type promotion semantics, and lots
> of errors were being thrown for trivial things like the arguments to pow().
> I fixed a bunch of these by hacking in casts, but eventually got bored before it
> finished.
>
> Now, if the C++ language standard was sane, I could just look up what the correct
> behavior should be and submit patches. But it isn't, so I won't. :)
>
I forgot about those. pow() caused a lot of errors. There are two
overloaded declarations, pow(float,float) and pow(double,double). We
have a lot of cases where is is called with pow(float,double). The new
Koenig lookup rules can't decide whether to promote the float to a
double or cast the double to a float. Our code is wrong and must be
fixed.
Bernie
_______________________________________________
Flightgear-users mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/flightgear-users
