- From: <noah_mendelsohn@us.ibm.com>
- Date: Tue, 3 Apr 2007 12:19:49 -0300
- To: ht@inf.ed.ac.uk (Henry S. Thompson)
- Cc: Dan Connolly <connolly@w3.org>, www-tag@w3.org
First of all, I agree with pretty much everything that Henry has written in his note. One refinement I would suggest: Henry Thompson writes: > What's important here is that consumers' wishes are paramount, but it > _is_ none-the-less possible for producers to state their wishes as > well. I don't think it's in all cases expressing a wish. Even in the typical case where the consumer has to make the ultimate decision on how to process information, how much validation to do, against what rules, etc. it can be very useful to allow the producer to indicate what rules he or she used in creating the content. For me, that's the fundamental reason for allowing language or specification identifiers. Sometimes it's worth the trouble, sometimes it's more confusing than helpful. Also: I think it's important to realize that having a single version identifier is in tension with allowing distributed extensibility. As long as you have a language that's evolved reasonably centrally, it's tractable for those involved to coordinate labeling of version information. When you have compound languages that allow piece parts to be invented by more or less anyone, then you may have HTML version x, with div #1 embedding an SVG version #2, or even in the worst case div#3 being a wrapper for some content originally authored by someone else in SVG version #1. It may even be that someone has just added a few local attributes to an otherwise standard HTML page. In such cases, trying to label a document with a single language version is at best a compromise, and at worst inappropriate. -------------------------------------- Noah Mendelsohn IBM Corporation One Rogers Street Cambridge, MA 02142 1-617-693-4036 --------------------------------------
Received on Tuesday, 3 April 2007 15:20:09 UTC