xml-dev - RE: Its the semantics dummy :-)

[johns@syscore.com (John F. Schlesinger)]

Sorry about the length of this post.

The argument over the infoset set me thinking about court reporters.

These are the people that record _verbatim_ the remarks made (by judge,
counsel and witnesses) during hearings (if that's the right term). Although
they report pauses and, sometimes, stutters, they don't report body
language. So arguably, they are reporting a 'reduced infoset'. The press
reporters at the hearing, if there are any, are reporting an even smaller
infoset. They are only reporting the sound bites (as opposed, I suppose, to
the sound bytes, of the court reporter) that will interest readers. In any
case, someone has made a decision as to what constitutes the information
content of the court hearing. The only way to report everything is to replay
(on the holodeck) the hearing. The legal system makes one decision, the
press another, about what it needs to report.

Similarly, someone has to make a decision about what to report about an XML
document. Reporting everything is much easier in the case of XML than a
court hearing. However, for the purpose of parsing the document into a
memory structure that can be processed by an application program, surely the
Infoset (or something like it) is what is needed?

If you want to be able to handle the entire document, you can still do that
yourself. If you object to a database mangling your document you can still
store it as text. But at least if there is an Infoset specification you can
know what to expect will and won't be mangled. Without the Infoset,
everything that processed the document would mangle it differently and you
wouldn't know what quirks of your syntax were safe.

Yours,
John F Schlesinger
SysCore Solutions
212 619 5200 x 219
917 886 5895 Mobile

-----Original Message-----
From: Simon St.Laurent [mailto:simonstl@simonstl.com]
Sent: Monday, July 31, 2000 5:54 PM
To: xml-dev@lists.xml.org
Subject: Re: Its the semantics dummy :-)


At 05:32 PM 7/31/00 -0400, Jonathan Borden wrote:
>Suppose we agree on MIME? If you send me a MIME document, I can process the
>characters using a MIME parser in 'MIME syntax' -- so what? It is always
>possible to send a perfectly well formed XML document that is totally
>useless e.g.
><doc> <byte>67</byte> <byte>121</byte>... </doc>

Uselessness is in the eye of the beholder, and MIME only takes you to the
doucment container, not to its contents.  67 121 might be a very important
code to me.

(Ever hear of those radio stations broadcasting only numbers?)

>The point is that some people might view XML as just a bunch of pointy
>brackets, and so what if the syntax is perfect *who cares*. Suppose your
>English syntax is perfect but you have nothing to say? What do I care about
>syntax?

If you have an important message to get to me, and you use German syntax, I
can promise that you'll be disappointed even if you use English vocabulary.

>The only reason we *need* agreed upon syntax is that we have something to
>say to eachother (i.e. semantics). What interests me about XML is the
>possibility of creating semantically meaningful documents.

That may be what interests you, but I find that outlook unnecessarily
limiting.
There are plenty of things I can do with XML as far as _figuring out_
semantics that I can't do with binary or otherwise encoded formats, and I'd
much rather you send me material in XML - even if I don't know the
semantics beforehand.

I'll take my syntax before my semantics, thank you.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
http://www.simonstl.com - XML essays and books