Padraig Bambury writes:
 > In his essay "A Technical Introduction to XML"
 > [] Norman Walsh writes:
Norm did write that quite a long time ago now.

 > over the web, some of the very features it lacks to make this
 > practical, make SGML a more satisfactory solution for the creation
 > and long-time storage of complex documents. In many organizations,
 > filtering SGML to XML will be the standard procedure for web
 > delivery".

I'd be rather surprised if Norm still said that so strongly, if at all

 > Could anyone please elaborate on what exactly these features are,
 > and why, if the statement is correct, SGML is "a more satisfactory
 > solution for the creation and long-time storage of complex
 > documents".

I see a number of areas you might want to consider:

 - for hand-authoring, omitted end tags, quote-less attributes, and
    short references are mighty handy. on the other hand, its easier
    just to not hand-author....
 - you may find features of XML DTDs irritating, viz the lack of
    inclusion and exclusion. but since we now have schema languages[1]
    which are even better, I dont see this as an issue
 - if the text is *massive*, omitted end tags could make a real
    difference in file size. something like the BNC, where practically
    every word is tagged, blows up in size alarmingly. do you care?
 - obscure SGML features like CONCUR might turn you on

you can see that I for one don't buy the pro-SGML argument. I think
the world has moved on from 1998's view of XML and SGML.

if someone can point me at *one* important new bit of SGML-only-aware
software which has come out in the last 3 years, I might reconsider!

Sebastian Rahtz

[1] when the final release of TEI P4 comes out (hopefully early in the
new year), it will come with a version in Relax NG schema form, for
those who like to experiment.