Re: XML for linguists?

On Mon, 22 Nov 1999, David G. Durand wrote:

>
> XML is like full SGML (not a tagset, but a full definition language with
> DTDs and everything), but it's had 90% of the cruft from SGML stripped out.
> SGML had a million complicating features intended to allow typing shortcuts
> that complicated the grammar unbelievably. It took most companies that
> built them at least 6 man-years to create conforming SGML parsers. Most XML
> parsers are created by single programmers in about a month and a half.
>
Moreover, it's really, really easy to work with XML, and gives a lot
of pleasure. It seems that inside SGML there was something elegant
and small struggling to get out... Of course, the fact that most
people have been getting used to open, text-based tagging languages
by using HTML has helped a lot, too, in the acceptance of XML.

> Not really. There are a lot of XML tools coming along, many free, but most
> are not quite end-user tools at this point. Especially in the document
> preparation area. Don't pin your hopes on the soon to be announced XML
> support in office 2000. It looks like it will be XML-complaint with a fixed
> DTD offering little more than RTF does.
KOffice (if you can get a snapshot to compile) already provides XML
output, but it's in the same league as Office 2000 - one type of
document. Perhaps the problem is that XML tags come in an unlimited
variety. I haven't looked into stylesheets or anything, but I'd doubt
that there is a standard mechanism that allows consecutive elements
to be represented in columns, like:

<e>string
  <t name="1">taggy</t>
  <t name="2>taggy 2</t>
</e>
<e>string2
  <t name="1">taggy 2'1</t>
  <t name="2>taggy 2'2</t>
</e>

 To be shown in the XML editor as:

 string   string2
 taggy    taggy 2'1
 taggy 2  taggy 2'2

I think that it's almost impossible to write a bit of software that
can accurately the intended presentation of all possible XML files, so
that we'll always need several specialized XML editors. (Like KWrite,
KIllustrator or KSpread. Or, Kura;-))

>
> The ISO created the HyTime standard based on full SGML. Steven DeRose and I
> wrote a book about it ("Making Hypermedia Work: A User's Guide to HyTime --
> maybe still available from Kluwer). HyTime was the source of some good
> ideas (but too off topic to continue here). I wouldn't pin my hopes on it
> though it is interesting to read about.
>
Hey, Kluwer is based in Deventer, where Irina and I live...

Boudewijn Rempt  | http://denden.conlang.org