Re: [COAS-List] COAS submission

[Tim Brinson <tbrinson@2ab.com>]

COAS Submitters,

The submission is impressive thanks to the hard work by Chuck, Tom, Larry
and everyone else working on it.  Dave Forslund's email about no
ObservationValues being defined is critical.  As I recall the Architecture
Board (AB) had explicitly requested the mandatory requirement about
defining an initial set of observation data types.  I figure these have
been planned but just not cfreated yet.

I did a quick review of the IDL for the service.  It looks quite good.  I
have a couple comments.

I'm not sure the ObservationData struct can contain the 'composite'
sequence of ObservationData.  I could be wrong here but as I recall valid
sequences have to be defined via typedefs and then used.  Some translators
may allow this but I don't think all will.  Maybe this has changed but I
think use to try this once and couldn't.  Up on looking at the IDL spec it
was not valid.  

The ramifications for this (assuming I am correct) is that you can not have
recursive structs.  In order to make it so you can create recursive structs
at runtime I have always put an "any" type into the struct.  I know of
other people that have used the same technique.

The other thing I notices is that EventSupplier and EventConsumer still
inherit from the Notification Service types.  At the Washington DC meeting
we discovered that by inheriting off the Event Service types someone could
use the Event Service or the Notification Service with it instead of just
the Notification Service.  I figure that is a change that has not been
completed yet but is still planned.

Chuck,

Around the middle of next week I will be spending some time with Pop at
Protocol Systems to work on the Waveforms IDL.  To do that I will need some
of the papers I had given you in Wash. DC.  Can you fax Pop the hand
written waveform IDL paper I gave you. If you also fax any other
ObservationValue hierarchy papers I might be able to creaet them too.


Regards,

Tim