Re: WWW/Internet 2009: 2nd CFP until 21 September
Date: Sat, 15 Aug 2009 16:38:11 +0000 (UTC)
Walter Mitty wrote:
>I have scanned the first few paragraphs of the article you cited. I admit
>that I haven't read it all, or any of it carefully.
>from the little I did read it seems clear to me that the originators of XML
>THINK that they have come up with a view of data that covers all of the same
>ground the the relational view of data covers. Or it's just possible that
>they never heard of the relational view of data. But just because they
>think that doesn't mean that you and I should think that.
All the author claims is that relational and XML-based representations map to each other. This is true; such mapping can be done automatically. But that doesn't mean the representations are equivalent: some things are easier to do in one representation than in the other. I do think the author should have given more prominence to the role of schema, query and data manipulation languages.
>So I return to my original question. Is XML simply a machine representation
>of data, or is it an alternative to the relational view of data? Another
>related question is, can you represent relational data in an XML document?
>Is anything gained or lost by doing so?
It is both a representation (the representation of trees as strings) and an alternative view (the trees themselves); see my other reply.
Of course relational data can be mapped to XML documents, with lossless, reversible mappings. (Make a database dump in XML format.)
What is gained is that you can send it over the wire and unpack it at the other end to create a copy of the database without having to worry about the tree-to-string serialization implementation being ill-defined or incorrectly implemented. The relation-to-labeled-tree mapping is simple in comparison. So that's what you gain. You only lose anything if you throw away the relational database management system.
-- ReinierReceived on Sat Aug 15 2009 - 11:38:11 CDT