Patrick,
We design some things in UML and some things in XML Schema here. It is not really that big a deal, especially when the goal is the exchange of metadata terms and/or queries. Schema does miss out in expressing certain inheritance rules, but we can keep track of these if they occur. There are a number of technical problems with working in a group with UML, but if everyone agrees to work on this in UML then fine lets do it.
But even in our OMG task force on an international standard for telemetry and commanding, we all came to the conclusion early on that it was most practical to develop with Schema and at the end convert to UML for OMG approval (OMG is the home of Model Driven Architecture). After we did convert to UML noone want to use it because it did not add anything and we did not have anything to add to it. For exchange languages (the L in VOQL) it is just is that way.
For UML, we would need to agree on an exchange language of the UML which may mean we need to agree on UML development environment. Right now the latest standard is XMI version 1.3, but not all UML viewers take that, or they take it only partially. One might be able to convert to GIF, but even a moderate size UML becomes too combersome to use this way. Then there is the issue of finding a good UML editor, since most except Rosetta are not that great either. Rosetta is very expensive. The practical fact is that XML has inexpensive and sophisticated schema editors that are very easy to learn.
As for your main point that we should not, right now, insist that the final VOQL be an XML language, I can go along with that. Who knows what is coming next and maybe we will all agree to use CORBA instead? But, as a practical matter, I think developing in XML and providing examples in XML is good enough for now.
Ed
Patrick Dowler wrote:
>On February 24, 2003 12:25, Kirk Borne wrote:
>
>
>>Ed's comments elaborate on the rationale for a VO Query Language,
>>but let's not forget the point ... VOQL is a language in a standard
>>format (XML) for specifying queries to distributed heterogeneous
>>astronomical information resources.
>>
>>
>
>Here I would like to strongly DISAGREE. There is no need to a-priori define
>VOQL in terms of XML. XML is a serialization format that one uses to
>exchange structured text. It is a huge mistake to pre-suppose that anything
>is fundamentally an XML thing. It is an even bigger mistake to "design with
>XML". I am not saying XML will or should not be part of the solution; it
>should not be goal...
>
>I suppose this comment could be addressed to every IVOA mailing list,
>but I'll leave it here for now.
>
>
>
Received on 2003-02-25Z18:49:09