- From: Daniel DuBois <ddubois@spyglass.com>
- Date: Wed, 15 May 1996 08:58:00 -0500
- To: Koen Holtman <koen@win.tue.nl>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
> - There is no spoofing mechanism: i.e. a response with come > Content-Location header may _not_ be used by 1.1 caches for > serving later requests which are done directly on the > Content-Location URI. I disagree strongly with this change, both in it's nature, and in it's timing. In my mind, such a change defeats the purpose of the Location/Content-Location header with regards to content negotiation, as it has been discussed and developed over the last 5-10+ months. As a process issue, there's little (no?) justification for the ground work laid down by the subgroups over many weeks/months to be tossed in favor of last minute changes by the editorial group. > - variant-IDs are deleted from the protocol. Their function is > taken over by the contents of the Content-Location field. Variant-IDs served the useful function of distinguishing non-name-able 'plain resources'. Content-Location served the more useful purposes of not only distinguishing but also identifying and locating name-able 'plain resources'. These simple, distinct premises fueled the entire Vary vs. Alternates, opaque vs. transparent, content negotiation structure. I'm not willing to see it fall out of the spec without significantly more convincing reasons other than a hallway discussion wanted to shrink the spec at the last minute. > This means that the entity tags of any two different entities > returned by different variant resources bound to the same generic > resource are guaranteed to be different. Another way to look at > this is to say that a variant-ID is opaquely encoded in the entity > identifier. I have no qualms with this as this does not affect what I perceive as the underlying principles of content negotiation as described above. Whether the variant ID information is encoded in a separate header is half a dozen of one, six of the other. > - Renaming `generic resources' to `negotiated resources' is > considered to be a good idea by some. > > - Renaming `entity tags' to `entity identifiers' is considered to be > a good idea by some. I agree with these, and have no problem with editorial terminology changes that are made for clarification purposes as long as they do not change the underlying mechanism. (Although I might have said "negotiable resources". "Negotiatied resources" sounds after the fact: in other words, that term could potentially be confused with plain resources, which means it's probably not much better.) I consider it a fundamental necessity that a "GET /index" that returns "/index.txt" should in every means possible be considered the same as a "GET /index.txt", including, especially, the abilty to be cached. When transparent negotiation 'really arrives' in the future, and people start manually requesting negotiated resources by their actual name rather than their generic resource URL, I don't want users to suffer any penalty, either in logic, treatment, or round trip times. In my dream future - web pages are going to be *chock full* of embeddable objects that can negotiated like we've never seen. Oh, you don't download images? Here's the HTML description of the image. Oh, you're blind? Here's an audio description of the images. Oh, you do java? Here's an annoying animation of the image. No one's going to want to tweak their Accept headers every time they come across a variant they want another version of. They're not going to be making their 'manual correct variant' request through the generic/negotiable resource URL, they'll be doing it through the URL they get from the transparent negotiation headers. ----- the Programmer formerly known as Dan http://www.spyglass.com/~ddubois/
Received on Wednesday, 15 May 1996 07:01:54 UTC