Re: [Autopilot] Conversion help.

[Trammell H. <hu...@sw...>]

On Wed, Aug 25, 2004 at 06:47:27PM -0500, Francisco Lobo wrote:
> So basically i can just replace the
> 
> >      const float             Pdot[2 * 2] = {
> >                         Q_angle - P[0][1] - P[1][0],    /* 0,0 */
> >                      - P[1][1],              /* 0,1 */
> >                      - P[1][1],              /* 1,0 */
> >              Q_gyro                          /* 1,1 */
> >      };
> 
> with
> 
> 
>         P[0][0] += Pdot[0] * dt;
>         P[0][1] += Pdot[1] * dt;
>         P[1][0] += Pdot[2] * dt;
>         P[1][1] += Pdot[3] * dt;

No, since the first computes Pdot using the current P, then updates
the covariance matrix P for the next time step.  You can, however,
replace Pdot[2*2] with Pdot[4], or, if you prefer the verbosity,
Pdot[2][2] and change initializer for the Pdot array to be a 2x2.

> Codevision does not like the Pdot[2*2].

That's a bug in their compiler.  2*2 is a compile time integral expression
and is valid as a size specifier for an array in C89 and C99 (and maybe
even in earlier pre-ANSI C).  This is the real reason I prefer gcc over
most embedded systems compilers -- gcc makes an attempt to adhere to
the standards, while most in-house compilers don't spend any effort on it.

Trammell
-- 
  -----|----- hu...@os...                   W 240-283-1700
*>=====[]L\   hu...@ro...                   M 505-463-1896
'     -'-`-   http://www.swcp.com/~hudson/                    KC5RNF