On Mon, 19 Apr 2010, Tim Jenness wrote:
> Does anyone know why ndfSbb takes a signed char but ndfQmask and ndfBb
> take an unsigned char for the badbit argument?
>
> The Fortran implementations (that are called) both say unsigned bytes in
> the description and both take simple bytes (since I don't think Fortran
> can do anything else) so I wonder why there is the discrepancy in the C
> wrapper.
Must be a bug. Looking at my NDF history (SCCS back to 1998) this code was
like this at the first commit and untouched since.
|