Re: XML character sets: a proposal

At 08:53 AM 9/13/96 -0500, Len Bullard wrote:
>I don't remember seeing that design criteria, James.  Being able 
>to do SGML by hand is a reason to do SGML.  Surprising numbers 
>of people do exactly this.  Again, we should not break existing 
>tools, among them, text editors.  Convenience is relative, of course, 
>but if XML is *difficult* to do by hand, I don't see that as a 
>strength.  That is where the heuristic of "convenience of
>implementation" 
>might have too much emphasis.

I was thinking the same thing. For better or for worse, most HTML markup is
still done by hand, and I would hope that XML would be useful and compelling
to the more advanced members of the HTML user community (even those who are
not now familiar with SGML).

Perhaps I could turn this around and ask: is ease of implementation really
as crucial as we have been treating it? I agree that it should be possible
to implement XML much more _efficiently_ than SGML, but I am skeptical that
a language that is "fanatically" easy-to-implement but difficult to write
data for will achieve more success than an author-centered language with a
few free, fast reference implementations.

Note though, that many of the SHORTTAG features actually make SGML harder to
write by confusing authors: "Why do I have to use quotes here, but not
here?" So I am not advocating that we preserve all (or even most) of the
keystroke saving devices in SGML.

 Paul Prescod

Received on Saturday, 14 September 1996 11:24:29 UTC