[moby] Re: [MOBY-dev] RFC #1863 Error Handling in MOBY-S -- Vote Called

[Mark Wilkinson]

On Wed, 2005-09-28 at 11:33 +0200, David González Pisano wrote:

> Using 
> the PIB also brings up again the problem of what is an empty object: can 
> we report back a "not so empty object", ie, an object with no data but 
> PIB data? What about an object with no data but CrossReferences? Is not 
> really clear to me. Given the examples about the PIB in the 
> specification, I always thought it was a good place to fill with static 
> data associated to my results (like software version, database release, 
> etc..), not really about why the service didn't work.

I agree with this interpretation of the API.  If a service fails, the
mobyData object should have the associated queryID filled in, but
otherwise have NO CONTENT.  This precludes the use of CrossRef's and/or
PIB's for error reporting.


> The last proposal  we uploaded includes a possible solution to avoid 
> XML-mixed elements, basically permiting serviceNotes to have two 
> elements: mobyException and Notes. This solution allows to extend the 
> servicNotes element with other uses/schemas in the future.

I think this is a great idea.  The serviceNotes block was stuck into the
original API just because it seemed so obviously useful, but was
intentionally left "undefined" until someone had time to think of how
best to structure it.  Looks like now someone has found the time!

M