Re: [galeon] WCS CF-netCDF profile document

Roy,

an important point you raise! At some point in our WCS discussion we
realized that actually a main difficulty is the variety a client has to
cope with: a server may choose to implement this or that functionality,
but a client should be interoperable to the max, hence needs to be
prepared for all variants.
Since then we rigorously cut out variants whenever possible. Actually,
in a perfect world the only variants we would have are the extension
packages - that is, the extension concept would formalize choice. A
client, then, upfront asks for the extensions implemented by the server
and decides whether it wants to continue conversation or not. Currently
we are working on a scheme that groups such extensions organically, as
one of our agenda items.

-Peter


Roy Mendelssohn wrote:
> Hi All:
>
> I think Jon Blower may have mentioned this earlier, but there needs to  
> be some differentiation in what is required of a server and a client.   
> While servers may be able to vary  (only 2-D or 3-D for example),  
> there must be a larger minimum set that a client has to implement in  
> order for there to be interoperability.  In code my lab is developing,  
> we are already running into this problem with WMS.
>
> -Roy
>
>