Re: Template URIs vs. URIs

On 10/5/06, Roy T. Fielding <fielding@gbiv.com> wrote:
> On Oct 5, 2006, at 5:40 PM, Mark Baker wrote:
>
> > As I mentioned, users could be confused about what's actually intended
> > to be a link in an email or a Web page.
> >
> > Spiders could be similarly confused, and as a result, waste time
> > chasing down links that aren't really links.
>
> In both cases, a correctly written parser will not do that (or will
> truncate the URI before the first open brace.

True, but IME (from observation, not implementation) URI detection
functions behave more like regexs than full parsers (for reasons of
efficiency, I expect), so you can't (IMO) fault them for false
positives on a string that looks like a URI save for "{" and "}".

>  An incorrectly written
> parser will do what you indicate regardless of the syntax chosen.

Unless the template syntax is sufficiently different from the URI
generic syntax, which was my suggestion.

> These templates are not going to be sent on email or appear in
> normal publications aside from the specification (except by accident),
> so there is no point in protecting against casual contact.
>
> All of the URI specs have specifically excluded braces from the
> syntax, partly because it was anticipated that we would need a
> syntax for variable substitution some day.  It makes perfect sense
> to use them for this purpose.  I am not sure if it makes sense to
> do this as an IETF standard, though, since I don't see this as a
> protocol element for Internet communications.

If that were the case, I would agree.  But the draft suggests that it
could be used as a protocol;

  "URI Templates can also be thought of as the basis of a machine-
   readable forms language; by allowing clients to form their own
   identifiers based on templates given to them by the URI's authority,
   it's possible to construct dynamic systems that use more of the URI
   than traditional HTML forms are able to."

Mark.

Received on Friday, 6 October 2006 14:20:20 UTC