W3C home > Mailing lists > Public > wai-xtech@w3.org > December 2004

Fw: Re: CG: XML (and SQL)

From: Will Pearson <will-pearson@tiscali.co.uk>
Date: Mon, 6 Dec 2004 14:23:32 -0000
Message-ID: <00cd01c4db9f$2e532360$c58b6051@WillPearson>
To: <wai-xtech@w3.org>

I thought the comments about TEI text would be interesting to some in pf.

Will
----- Original Message ----- 
From: "Simon Polovina" <S.Polovina@shu.ac.uk>
To: <will-pearson@tiscali.co.uk>
Sent: Monday, December 06, 2004 2:15 PM
Subject: FW: Re: CG: XML (and SQL)


> Will fyi - Simon
> 
> -----Original Message-----
> From:  John F. Sowa
> Date:  6/12/04 2:03
> To:  cg@cs.uah.edu
> Subj:  Re: CG: XML (and SQL)
> 
> Peter,
> 
> Thanks for the extra data point.  It's always
> nice to have a couple of factoids to leaven
> the speculation.
> 
> PL> Despite John's fears, sometimes the right
>  > technology decisions are made by the industry
>  > - and relational databases are much more right
>  > than hierarchical or network databases...
> 
> I believe that good technology tends to win in the
> long run, although there are many speed bumps along
> the way.  The marketplace works fairly well, but
> the greatest hindrance to an efficient market is a
> monopoly or cartel.  Sometimes the effect of a cartel
> can be created by a fad or a premature standard.
> 
> Re relational DB:  I was a strong proponent of
> RDBMS since the 1970s, and my major complaint
> about SQL was that it wasn't as good as it should
> have been.  But it's still better than most of
> the alternatives.
> 
> PL> Of course, the original engineer did no such
>  > modelling and just blindly kept the XML structure
>  > and replicated it through the system. My rough
>  > estimate is that this has resulted in at least 3x
>  > more development and maintenance effort than if
>  > things had been properly modelled. The fault is
>  > not in using XML as such - it's in being seduced
>  > into wrong design decisions because of XML.
> 
> That's a very good point:  XML is more than just
> syntax, because it has a strong influence on the
> data model.  There's more to semantics than data
> modeling, but the data model is more fundamental
> than the surface syntax, since it has tentacles
> that reach deeper into the semantics.
> 
> I also received another offline note about the Text
> Encoding Initiative (TEI), which started to use SGML
> and later XML for annotating natural language texts.
> It turns out that the users of TEI texts discovered
> that putting annotations in texts creates problems:
> it only allows one perspective on a text, but there
> are often many different, equally valid perspectives
> on the same text for different purposes.
> 
> For this reason, many people are beginning to think
> that combining metadata with the data is not necessarily
> a good idea.  In fact, it might be better to separate
> the data from the metadata in order to have multiple
> metalevel interpretations or annotations of the same
> source from different perspectives for different purposes.
> 
> As I said before, I think XML is very useful for many
> purposes, and I suggested it to David Cox for FLIPP.
> But it's not a panacea, and it's not a good idea to
> edict it for every application.  You have to think
> carefully about whether it should or should not be
> used on a case-by-case basis.
> 
> John
> 
Received on Monday, 6 December 2004 14:21:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:15:39 GMT