Re: [Fwd: [COAS-List] Potential Issue for the FTF]

[David Forslund <dwf@lanl.gov>]

At 02:36 PM 7/1/99 -0700, Tim Brinson wrote:
OK, but we need to get a little more specific.  Does the proposed 
StructuredObservationData have
ObservationData as its elements or StructuredObservationData?   Is this 
change also proposed
to get around the marshalling problem you mentioned in your email, so that 
it has ObservationData[]
as its component to avoid the recursion?   Is ObservationData then a 
[name,value <any>] pair or just an Any?

It seems then that the queries would be through an Any, too?

We need to sit down and work out the example structures that COAS would 
support with this model.

Fundamentally, it seems to me that this is more than a "simple" change to 
COAS, but I guess I need to see more
specifics.

Thanks,

Dave


>Dave,
>
>The attached email is the one I sent about the COAS issue.  The solution
>would be to rename the ObservationData structure to
>StructuredObservationData and to create an ObservationData type that is
>typedefed as an "any".  That way you can send anything as the
>ObservationData type (even ObservationRemote references).  You can still
>send the structure StructuredObservationData through the "any".
>
>Over time we may find other more fixed structures that are commonly
>needed and be able to standardize them.  They would still be able to be
>passed via COAS.
>
>Let me know how it goes.
>
>Tim