[CF-metadata] missing_value vs. _FillValue

[John Caron]

Brian Eaton wrote:

>Hi John,
>
>On Thu, Nov 06, 2003 at 10:11:12AM -0700, John Caron wrote:
>  
>
>>I dont think of  _FillValue as defining the valid range, but rather its 
>>one specific value that is reccomended to be outside the valid range if 
>>that is defined.
>>    
>>
>
>The NUG (section 8.1) under the entry for valid_range describes how to use
>the _FillValue attribute to define the valid range if it hasn't already
>been defined by the valid_range, valid_min, or valid_max attributes.
>Since _FillValue has default values this means that the valid range also
>has default values.
>
>Brian
>_______________________________________________
>CF-metadata mailing list
>CF-metadata at cgd.ucar.edu
>http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
>  
>
 _FillValue creating a range definition seems to me to be an unexpected 
side-effect of pre-filling data. Using the default _FillValue seems 
really wrong, since theres no way to turn it off. I would say only if 
_FillValue is explicitly defined should it define a missing value.

I'll have to think more about the range thing, i wonder if its motivated 
by the problem of exact comparision of floating numbers ?