Re: The document's address

[Albert Lunde <[e-mail]>]

On 3/8/2013 10:10 AM, Nicholas Shanks wrote:
> On 8 March 2013 15:42, Julian Reschke <[e-mail]> wrote:
>
>> That implies a concept of hierarchical ownership that simply does not exist
>> in HTTP. It might for some servers, but there's no guarantee.
>
> Can you provide an example (make one up) where ownership cannot be
> defined as hierarchical and accumulative?
>
> I am suggesting that HTTP's concept of "ownership" (for purposes of
> replacing cache entries) be defined by the specs to be hierarchical,
> since if I own / and I want to sabotage /subdir/ all I have to do is
> log in to the server and replace/delete it.

That's only true for a server based on a simple file system with no 
aliases.  I used to run a Apache web server where the home page and the 
top-level menu content were all aliased in from another directory, and 
htdocs was owned by root and rarely changed. This was to prevent things 
like Dreamweaver from nuking unrelated content.

But this is an implementation detail, unrelated to what's on the wire; a 
dynamic database-driven site, like a Web CMS, can implement completely 
different scopes of control.