[Gdal-dev] Re: Nodata bitmasks / null processing

[Ray Gardener]

Tim Keitt wrote:

> I would require that a band with a mask must not have a nodata
> value (make them orthogonal).

If the goal is abstraction, why not move to the masking API since it is 
more scaleable? This makes it possible to also upgrade to confidence 
values, e.g., values indicating how likely a pixel is to have the 
assigned value (or error tolerance indicators). For now, use a zero mask 
value to indicate 100% void (0% confidence).

Ray