Re: What is the logic of storing XML in a Database?

From: Marshall <>
Date: 30 Mar 2007 09:46:18 -0700
Message-ID: <>

On Mar 30, 8:27 am, Bernard Peek <> wrote:
> On 2007-03-30, Marshall <> wrote:
> > That is a fair point. However I don't buy the rest of it. Pushing
> > binary data was a solved problem thirty years ago. There was
> > no standard for arbitrarily structured binary data, and it turns
> > out that arbitrarily structured data is really useful. It so happens
> > that the first widely adopted standard for arbitrarily structured
> > data was XML. That's the only idea of value that I have identified
> > so far in XML. It's a very good one, but the large number
> > of bad design choices (such as using a text representation,)
> > the poor efficiency (which in part results from using a text
> > representation) and the unnecessary complexity make it
> > otherwise unattractive.
> I can see why a text format was adopted, I'm sure that there were lots of
> people clinging to the idea that the data ought to be editable in vi if
> necessary. I happen to think that editing data files by hand is not a smart
> thing to do, because it encourages people to fix symptoms instead of causes.


I notice that there is already a well understood domain in which text formats are convented into binary in ways that are potentially quite complicated and yet has been mastered by a large body of programmers, and also has a mature error handling process. I'm speaking of turning source code into object code.

These techniques are really, really well understood, and work really well. They should be applied to data.

> Thankfully XML developed sufficient complexity to effectively discourage that.

Hrm. It seems you and I have different values.

> The efficiency argument I've already covered, but I should add that it's
> normal for XML files to be compressed, which effectively turns it into an
> efficient binary format.

Compression addresses the bandwidth costs, but not the computational overhead. You still have to uncompress and parse, and this is not cheap.

Marshall Received on Fri Mar 30 2007 - 18:46:18 CEST

Original text of this message