RE: ACTION- 541: Jeni to help Dan pull together terminology on Deep Linking

The generic distinction between cache and archive is much more a distinction in function, and the critical element for the "deep linking" discussion revolves around the ordinary usage as it relates to copyright and security:  a "cache" as a transient copy implemented merely for performance optimization should be expected to work as if you were accessing the original resource...   a properly implemented cache does not introduce different access control rights, and should even not introduce a different lifetime. (That is, a resource that is there and goes away is one where the access control is originally wide and then reduces to zero.)

On the other hand, an "archive" is that it is explicitly intended to have a different lifetime than the original resource. Even if the owner of the original resource explicitly removes the material from access, it might well be retained by the archive.   It makes sense to archive access controlled material and (at some later date) make it available to a different set of principles than the original resource.

Whether or not the resource content is stored and accessed via the same or a separate URL is orthogonal. You can imagine an archive which uses the same URL as the original (as memento proposes, I believe), and it is easy to imagine a cache which rewrites URLs to redirect retrieval to a content distribution network.

Larry
--
http://larry.masinter.net


-----Original Message-----
From: www-tag-request@w3.org [mailto:www-tag-request@w3.org] On Behalf Of Chris Lilley
Sent: Thursday, April 07, 2011 6:16 PM
To: Noah Mendelsohn
Cc: Henry S. Thompson; Jeni Tennison; Dan Appelquist; www-tag@w3.org
Subject: Re: ACTION- 541: Jeni to help Dan pull together terminology on Deep Linking

On Thursday, April 7, 2011, 7:32:06 PM, Noah wrote:

NM> On 3/31/2011 1:51 PM, Henry S. Thompson wrote:
>>    *cache*: to store a copy of something hosted elsewhere and
>>    likewise make it available

NM> I feel like this is missing something along the lines of:

NM> *cache*: to store a copy of something hosted elsewhere and likewise make it
NM> available. Cached copies are typically created to improve performance or
NM> availability, and are usually not managed for long-term stability.

It also seems to be missing the distinction that a cached copy is the same resource; the URI does not change, there just happens to be a copy stored along the network path.

A cached copy thus differs from a copy which is served from a different URI (which may be an out of date snapshot, either inadvertently or deliberately, such as with the wayback machine).

NM> I don't love that, but I feel your original definition misses the point of
NM> a cached copy: it's typically an optimization, and crucially, nothing in
NM> the system should behave differently if that copy disappears, except for
NM> perhaps being slower or less able to respond in the face of network partition.

NM> Noah




-- 
 Chris Lilley   Technical Director, Interaction Domain                 
 W3C Graphics Activity Lead, Fonts Activity Lead
 Co-Chair, W3C Hypertext CG
 Member, CSS, WebFonts, SVG Working Groups

Received on Friday, 8 April 2011 18:23:31 UTC