xml:id RE: [sc34wg3] Compact syntax requirement question
Lars Marius Garshol
Wed, 20 Jul 2005 16:48:26 +0200
* James David Mason
| I think that had we had, a quarter of a century ago, [...]
A lot of good points there, Jim. I'm going to just let them pass
without comment, I think.
| I'm generally opposed to standardizing new syntaxes unless there are
| technical reasons why it is unavoidable. Verbosity is not a
| technical reason. For this reason, I have very little enthusiasm
| for CTM. I likewise have little enthusiasm for a TMQL that is not in
| XML syntax.
I really don't think a TMQL in XML syntax is going to be usable. Of
course, it's possible to write queries in XML, and some people have
even defined query languages with XML syntax, but if we want some
adoption for TMQL (and one major reason to create TMQL is to make the
Topic Maps package as a whole more compelling for users) we're going
to have to create a more developer-friendly syntax.
Note that even XPath and XQuery do not have an XML syntax. Nobody
seems to be bothered by this, but when XPath did have an XML syntax
(the first XSL draft, done by James Clark and somebody from
Microsoft), people *were* bothered to the extent that it was just
quietly dropped and replaced by today's XPath.
So, really, I think the only choice here is whether there should be an
XML syntax of TMQL in addition to the non-XML syntax. Personally, I
don't see the point of that.
| If you can't notate it in XML, is it even in the scope of SC34?
I don't think SC34 is only about XML; if it were then topic maps
wouldn't belong in SC34.
| Verbosity is not in general an excuse for not using XML. I will
| readily admit that XSLT is hideously verbose, and I paid good money
| for an editor that speeds up creation of the stuff. But I'll put up
| with that. (I didn't learn DSSSL because I didn't want to branch out
| into Scheme, and I learned only enough OmniMark to make small
| changes in others' code. But I learned XSLT willingly.)
Verbosity is precisely the reason, I'm afraid. For a query language
(and a syntax like CTM) it is really important that the language isn't
wantonly verbose, because that is going to drastically reduce the
usability of the language in practice.
Your point about XSLT is well taken, since XSLT with an XML syntax
really has worked quite well, but then XSLT is a special case. Its
primary purpose is to create XML, so the developer is already writing
XML. Secondly, a major part of XSLT (XPath) does *not* have an XML
syntax, and most of the heavy lifting is actually done in XPath. (See
above for what happened when XPath was in XML, too...)
Note also how XQuery quite deliberately chose to go the other way and
use a non-XML syntax. This despite the experience with XSLT.
| So I wonder whether CTM is really worth SC34's time.
Given what I wrote about TMQL above, I guess it's obvious why I think
we should have CTM (given what I wrote in reply to Bernard's
Lars Marius Garshol, Ontopian <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50 <URL: http://www.garshol.priv.no >