[phplib] why not XML/LDAP From: fred <email protected>
Date: 08/05/01

On Sun, Aug 05, 2001 at 07:22:17PM +0200, Kristian Koehntopp wrote:
> See, with the invention of LDAP and XML, people are falling all
> over themselves to throw away their hard earned relational
> database knowhow and go back to hierarchical structures. They
> are chasing buzzwords, and this is going to cost them.

What follows is a nice analysis of how XML is not suitable as the
"leading" (authoritative) storage representation for many
applications.

> and they [XML and LDAP] have only weak integrity checking and weak
> data types.

With XML Schema, XML data values can be constrained at least as well
as the datatypes and checks do in SQL. However, I don't think XML
Schema addresses referential integrity.

> Oracle is lacking such a tool [as mysqldump] in the default install,
> so you can put data into an Oracle database, but you cannot get data
> out of an Oracle base in a self-descriptive form, nor can you get a
> reuseable schema description from an Oracle.

I was amazed and dismayed when I first encountered that, but it seems
endemic to market-leading closed source software. Try exporting
useful data from QuickBooks, for example. I've heard that the UCITA
allows software vendors to lock up our data even tighter, restricting
our right to reverse-engineer proprietary data formats to get our data
back out again.

PostgreSQL also has good facilities for dumping an entire database.
And since the current PostgreSQL features for altering a schema on
the fly are limited, it's sometimes necessary to dump, hack schema,
and reload as you describe.

> But, does an application want to store its internal data in XML
> representation?
>
> Certainly not.

That seems too strong if we include applications like document/content
management -- and why wouldn't we, on this forum. I've found XML
markup to be very productive as a base representation for documents
that are then rendered as XHTML. Representating those same documents
as the join of fragments from a relational database, no matter how
elegant the tree/relational mapping, sounds like a major hassle. If
there _is_ some practical way to do that, I'd like to know how. I
don't mean building a web page from parts that come from a database --
I do that all the time -- I mean the semantic markup of narrative (but
structured) text, such as is often done based on the DocBook schema.

Overall, though, your argument for XML's role is well-stated and
convincing. Thank you, Kristian, for providing it. 

-- 
Fred Yankowski           fred <email protected>      tel: +1.630.879.1312
Principal Consultant     www.OntoSys.com       fax: +1.630.879.1370
OntoSys, Inc             38W242 Deerpath Rd, Batavia, IL 60510, USA

-- 
Abbestellen mit Mail an:   phplib-unsubscribe <email protected>
Kommandoliste mit Mail an: phplib-help <email protected>