RE: [metaDataInURI-31] New draft of "The Use of Metadata in URIs"

Hi Noah,
Nice work, but I wonder if this "Good practice" note is really adding
Good Practice: Web sites SHOULD use hyperlinks and URI metadata in a
manner that minimizes confusion for users, and SHOULD NOT misleadingly
apply common conventions for encoding type information into filenames
and URIs.

Isn't it rather like saying Web sites SHOULD be designed well and SHOULD
NOT do malicious things?  It seems like kind of generic advice.

David Booth, Ph.D.
HP Software
Phone: +1 617 629 8881



	From: [] On
Behalf Of
	Sent: Tuesday, November 07, 2006 12:28 PM
	Subject: [metaDataInURI-31] New draft of "The Use of Metadata in

	I am pleased to, somewhat belatedly, make available another
draft of a TAG finding on The use of Metadata in URIs [1,2].  An earlier
draft [3] was reviewed at the TAG's Face to Face meeting in Vancouver in
October of this year.  Most of that draft was considered ready for
publication, but a few changes were suggested (see F2F minutes at [4]).
The draft published today is intended to address the changes requested.

	Most of the changes were straightforward, including: 
	* Removing the good practice note from section 2.2 
	* Changing several occurrences of the word "identify" to URI 
	* Fix Stuart Williams' email address in the front matter 
	* Fix various typos 
	The substantial change requested involves section 2.8, which had
been titled "Malicious metadata".  Comments from TAG members on the
original [5]          included statements to the effect of "that's
exactly backwards from the story we want to tell" (constructive feedback
is always welcome), and "we want to talk about the general confusion
users have about the mappings from URIs and served media types to
platform file naming conventions".  There was also a sense that we want
to make clear that when a file is saved from the Web to a local OS, the
appropriate default mappings should be from the authoritative media type
of the served representation to suitable equivalents on the platform.
Other correspondents on this mailing list expressed similar concerns. 
	A couple of weeks ago I sent an email asking for clarification
of some of TAG members' concerns [6], but having received no responses,
I've gone ahead and redrafted based on my own understanding.  If this
requires further rework, that's not a problem from my point of view.  Of
course, I'd like to believe that we're (finally) getting ready to
publish this one.   So, the new draft includes a complete rework of
section 2.8 [7].  It bears the new title "Confusing or malicious
metadata".  I only wrapped this up a few minutes ago, and it needs some
proofreading. I think it's in good enough shape to be worth reviewing. 
	Please accept my apologies for sending this so shortly before
today's call.  Maybe it will prove simple enough to discuss on short
notice, or else we can discuss next week after TAG members have had time
to read it.   In any case, you can quite safely jump to the new section
2.8 [7] if you're attempting a quick read before the call.  The other
changes are low risk, I think. 
	Noah Mendelsohn 
	IBM Corporation
	One Rogers Street
	Cambridge, MA 02142

Received on Friday, 10 November 2006 03:29:12 UTC