RE: typing (was RE: Personal reply)

["Simon St.Laurent" <simonstl@simonstl.com>]

At 03:48 PM 3/12/01 -0500, David E. Cleary wrote:
>If I have a service on the Internet where I say I support messages or
>documents conforming to a particular schema, then anyone who wants to send
>me something has to agree that value is a decimal number. If they send me 
>an alphanumeric string, I will not validate it and reject it. Even without
>schema's and using a DTD, my application would rightfully reject it even
>though that instance is valid in regards to the DTD. As far as software 
>that accepts arbitrary XML markup and is able to generate usefull work 
>from that, I'll leave that to the acedemics. It is not something that adds 
>any value to my areas of interest.

The use case you're proposing here doesn't actually deviate in the least 
from what I said earlier, because you _are_ the recipient.  As the 
recipient of messages, you have the power to accept and reject requests.

Do senders have the same privilege?  No.  This isn't "academic", something 
wacky to be contrasted with "common sense".

>Stick with DTDs, DOM Level 1 or 2, and SAX and you are all set. However, 
>you shouldn't stand in they way of those important set of use cases from 
>taking advantage of datatyping in XML.

I'll stand in the way of XSLT and XPath and XForms requiring the use of XML 
Schema to the extent that XML Schema usage is an impediment to my using 
those specs.  Seems pretty reasonable to me.  (Especially given the earlier 
language in XForms about the "hearts and minds of developers".)

>Sorry, but since the W3C created XML, I would be surprised if XML Schema
>wasn't priveledged in this regard.

I'd take that more seriously if more W3C participants (not just large 
vendors, thanks!) showed a unified front on the value of schemas and how 
good this spec is.  Mostly, I see people holding their nose, and some up 
and leaving.

>The only problem I see being addressed here is that some people don't like
>XML Schema and would rather see it replaced with their own favorite schema
>dialect, or nothing at all.

So it's unreasonable to dislike XML Schema for a wide-range of technical 
and political reasons?  Is there something deeply wrong with trying to do a 
better job?

>The way to move forward is to start making the
>PSVI more suitable for any schema dialect, not throwing out XML Schema. It
>should be possible to take a well formed document with no schema or DTD 
>and generate a default PSVI that allows it to work with PSVI aware APIs. 
>That is much more usefull than throwing darts at the W3C or XML Schema. 
>But then again, you have those who argue the Infoset, let alone a PSVI, is 
>bad for XML.

The PSVI looks like some pretty ugly stuff to me.  On the other hand, I'd 
rather have a publicly-defined PSVI with a clear set of explanations for 
what it's supposed to be about than attempting to summon the spirit of PSVI 
through repeated readings of the XML Schema specs...


Simon St.Laurent - Associate Editor, O'Reilly and Associates
XML Elements of Style / XML: A Primer, 2nd Ed.
XHTML: Migrating Toward XML
http://www.simonstl.com - XML essays and books