[Standards-JIG] JEP-0060: yet another 1.8 draft

Bob Wyman bob at wyman.us
Mon Feb 6 23:37:05 UTC 2006

I'm a bit concerned about the fact that you've actually gone ahead and
defined syntax for content-based filters... Particularly a syntax which is
as weak as what seems to be implied by your example of 'pubsub#keyword'...
If you're going to follow this path, then it would probably make sense to
define something which is actually useful... Given today's tendencies, it
would seem that what we need is something that relies on XPath or a subset
of XQuery. Also, since PubSub.com currently implements such a syntax, does
this mean that we should get our syntax registered?

In the case of pubsub#keyword:
	If the XML stanza that this element is applied to contains more than
one element (for instance, both a title and a description), which element is
used for the match?
	Can more than one keyword be specified in a pubsub#keyword field?
	If the XML Schema type of an element which is to be matched against
is defined, then would normalization of keywords be permitted in matching?
For instance, if the interesting element is a floating point number, would
"1.0" and "1" match? If the type of the field is "date" and the current date
is 6-Feb, would "yesterday" match against a date of 5-Feb?
	It should be noted that keywords will often contain characters that
cannot be passed in well-formed XML without difficulty. Thus,
implementations should be prepared to parse CDATA.

	bob wyman

More information about the Standards mailing list