On Tue, Apr 29, 2008 at 1:59 PM, Andrew S. Townley <[EMAIL PROTECTED]> wrote: > I see it that the client should save a reference to the current resource > if it finds that resulting resource "interesting", but it shouldn't know > that it can retrieve that resource again using some kind of template > operation to construct a URI. That's what I meant about "URIs only > being meaningful to whomever resolves them". Certain layers of the > client may need to resolve parts of the URI, e.g. the scheme, host and > port in the case of HTTP, but that's as far as it should go, because > that's the only part that's meaningful to the client according to the > published specification.
While I agree with many of your sentiments overall, I think your strict prohibition against URI templates goes too far. I posted my reply as a comment to your blog post on URI opacity<http://atownley.org/2008/04/uri-opacity-revisited/>. I also posted my reply in the REST discussion group<http://tech.groups.yahoo.com/group/rest-discuss/message/10669>. I think that is the more appropriate forum for this issue. I look forward to continuing the insightful discussion there. -- Nick
