On Mon, 14 Apr 2008, Malcolm J. Currie wrote:
> It did alarm me to see a subroutine call for every pixel in the existing
> code.
Here's an unrelated one that might alarm you: Why does it take longer to
use COPYBAD to copy bad pixels from one NDF to another than it does to add
those two NDFs together using ADD? I discovered this when doing bad-pixel
masking for large ACSIS cubes. COPYBAD runs about 30% longer than ADD.
I suspect there might be other things like this lurking in the Starlink
code tree...
Cheers,
Brad.
|