Re: XML versus Relational Database

[Michael Beddow <mbnospam@mbeddow.net>]

I agree with everything in Soumitra Sengupta's posting. But let me add
a more general "it depends". It depends, too, on whether you see your
project as being the sole interested party in your data, now and in
the future.

If that's so, you need consider nothing beyond the technical and
business criteria when deciding to go xml or not.

But I'm interested in scholarly texts and lexicographical data about
natural languages. With the first category, virtually everything
points towards xml anyway. But lexicography is different. If authors
of electronic dictionaries want tried and tested technologies that
allow complex categorisation and retrieval of their specific data
quckly and efficiently, at the moment they probably ought to go for
RDMBS or OODMBS solutions (or proprietary sgml-based ones.)

But if they do that, they are locking their data into a specific (tho'
not necessarily proprietary) format with specific retrieval
requirements. If instead they opt for XML, they leave it open to
others easily to access their data and merge it into their different
but related projects, even ones that haven't been thought of yet.
Certainly in the field of scholarship, that hugely enabling potential
of XML seems to me to outweigh any loss of immediate speed or
efficiency that  conventional database solutions might offer.

Michael
------------------------------------------
Michael Beddow
http://www.mbeddow.net/