OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: XML versus Relational Database



Title: RE: XML versus Relational Database
Nothing.  I am saying any of that can be had pretty much
anytime with commercially available systems.  Today,
XML and databases are embarassingly surfeit with
capability and it becomes a problem of choosing well.
 
On the other hand, we have been handling the problems
of heterogeneity and transform well without XML or
XSLT for quite some time now.  It is expensive
and one-off, but XML and XSLT only bring down
the complexity of that.  The need to negotiate,
analyse, plan, propose and resolve are still
the expensive bits and it is still noisy.   The problem
is one mans metadata IS another mans data
and the differences are what take up all the time
and profit.  You are glossing across that "semantically
equivalent" problem too easily.  Technology is not
the issue there.  Locality is.

Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

-----Original Message-----
From: Linda Grimaldi [mailto:grimlinda@earthlink.net]

I guess I don't quite understand the statement that "XML vs Relational DB is a non-issue".  If your needs are extremely data-centric and you don't much care about extensibility- the info you store does not change structure- then I suppose there is truth to it.  However, for heterogeneous stores of information from multiple sources, many of whom, despite best efforts, may not conform to a standardized schema and yet may be providing semantically equivalent data, it seems to me that an RDBMS solution, with or without embedded XSLT, is pretty much inadequate, mostly due to the need for metadata predefinition in an RDBMS and for performance issues around transformation.
 
XML allows metadata and data to be treated symetrically, with context dependent entirely on each query. One man's metadata is, after all, another man's data. RDBMs, by their very nature, strike me as quite limited in this regard.  What am I missing?