Thomas McGlynn wrote:
> Ed Shaya wrote:
>
>> Personally, I think it would be good idea, when designing a new XML
>> format, to allow the mask and the data be combined by allowing
>> special numbers assigned to the different Nulls.
>
>
> If you are intending that the different NULLs use different NaN bit
> patterns,
> one thing that will cause problems is that Java converts
> all NaNs to a single fiducial bit pattern in any manipulation of the
> data.
I do not advocate using binary!
In ADC/CDS tables, authors have often chosen the lowest available numbers
within the Format statement like -999 and -998 or -99.9 to represent the
Unknown values.
Often, to my dismay, they have chosen 0.00 to mean null.
>
> Personally I think we use masks not because we have too few choices
> of nulls, but because masks do not throw out whatever data we have
> in the image and can express criteria (not just badness) about our
> data with far more flexibility.
>
But even greater flexibility is to allow for metadata on null values in
the data OR to allow a mask. Within the context of VOTable, a mask is
not usually the preferred method (but we could enforce it).
Either way, one would want to make it (nullValue and/or masks) part of
the standard vocabulary so that applications understand it and behave
properly.
Ed
> Tom
Received on 2006-06-23Z19:51:00