Re: Rolled up Feedback on draft-gregorio-uritemplate-00

On Nov 9, 2006, at 7:54 PM, Roy T. Fielding wrote:

>
> On Nov 9, 2006, at 3:19 PM, Joe Gregorio wrote:
>> On 11/9/06, Marc Hadley <Marc.Hadley@sun.com> wrote:
>>> I also got some feedback on the users@wadl.dev.java.net mailing list
>>> that the spec should require at least one reserved character between
>>> template variables since, in the general case, its impossible to  
>>> work
>>> out the values of the template variables after substitution. E.g.
>>> given the template
>>>
>>> http://example.com/{foo}{bar}
>>>
>>> and the URI
>>>
>>> http://example.com/xyzzy
>>>
>>> you can't tell what the values of foo and bar are.
>>
>> I didn't think that was a use case. I have been operating
>> under the assumption that we were covering
>> URI Template -> URI, but not the other direction.
>
> It isn't a use case. In general, it is just a bad idea.
>
> Templates should be allowing an owner to tell the client how to
> construct a URI.  Trying to go the other way means the client is
> attempting to understand the URI by searching for things that
> look like metadata.  That is a well-known bad idea, violating
> several findings and Web design notes.
>
Actually the use case was server side. The commentator wanted to use  
URI templates to deploy code that would be executed when a request  
was made on a matching URI.

Marc.

---
Marc Hadley <marc.hadley at sun.com>
CTO Office, Sun Microsystems.