Re: [xml-dev] What does it mean to say that XML was over-engineered?

[Peter Flynn <peter@silmaril.ie>]

On 17/09/2021 12:51, Elliotte Rusty Harold wrote:
> DTDs don't really bother me. Internal DTD subsets are where the 
> problems arise. If we had stuck to external DTDs that were only 
> optionally processed and predefined all the HTML entity references,
> a lot of problems would go away. Also minimal parsers would be much 
> simpler and possibly faster.  Processing instructions maybe aren't 
> needed but don't cause a huge amount of problems.

I come from the other end. DTDs are critical infrastructure for many
publishers, and subsets of the structure are used by invoking a spell in
the internal subset. None of the ones I deal with use any entity
references these days other than the five defined for XML, with a tiny
handful of exceptions. PIs are very much needed, but I am happy to hear
they cause no problems for others.

> Namespaces are necessary but almost certainly could have been done
> better.

The first time I heard about them I thought they were going to do
something quite different¹. I don't have a particular problem with them:
when using someone else's schema they're basically just a tedious excise.

Peter
-- 
¹ What I had imagined — very short-lived — was that the URI of a
namespace would resolve to a schema from which the declaration of an
element type would be taken with all its potential children, so that I
could construct my new schema with OL|UL|DL "done as they are in HTML",
BLOCKQUOTE|FIGURE|TABLE "done as they are in DocBook", bibliographies
and references from TEI, metadata from JATS, etc, etc, ad
confundationem. Mercifully the days of writing new schemas and DTDs is
largely over.