Re: null allowed in range of Fields

Hi Andrew,

Thanks for finding this. The fix is on the VisAD ftp server.

Cheers,
Bill

On Thu, 15 May 2003, Andrew Donaldson wrote:

> We are storing radar data with a function of:
> Time-> ((range, azimuth)->reflection)
>
> Sometimes there is no reflection from a radar - when it's not raining.
> It is important that the display shows no rain when there is no
> rain, including when visad is animating and there is rain on the
> preceding and next sample.
>
> I have been experimenting with storing:
> Time -> null
> in the field.
>
> This works when I use setSamples() but fails when I use setSample(int,data).
>
> Looking at the code for FieldImpl, it seems that null is allowed
> when using setSamples() as it is tested for on line 209 and assigned as
> a literal on line 218.
>
> Looking at setSample(int, data) line 677 does not do this test for
> null, but throws an exception when evaluating range.getType().
>
> This seems to be inconsistent.
>
> It looks like Bill purposely allowed null with his code change
> on 9 Dec 1999 line 208, so I think null is intended to be supported.
> I hope it is.  If so, then a small change to setSample(int, data) is
> needed.
>
> If not, then how should I store the data for my case above?
>
> thanks,
>
> Andrew.
>
> --
>   Andrew Donaldson
>   Email:   A.Donaldson@xxxxxxxxxx                Bureau of Meteorology
>   Phone:   +61 3 9669 4537            P.O. Box. 1289K/150 Lonsdale St,
>   Fax:     +61 3 9669 4128                       Melbourne, Australia.