W3C home > Mailing lists > Public > uri@w3.org > October 2007

Re: file: URIs without host

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Mon, 15 Oct 2007 11:26:57 +0100
Message-ID: <47134071.6090300@hpl.hp.com>
To: Charles Lindsey <chl@clerew.man.ac.uk>
CC: URI <uri@w3.org>

Charles Lindsey wrote:
> 
> On Thu, 11 Oct 2007 01:44:18 +0100, Daniel R. Tobias <dan@tobias.name> 
> wrote:
> 
>> On 10 Oct 2007 at 20:45, Charles Lindsey wrote:
>>
>>> It is only the rare cases where you want to change scheme at the same 
>>> time
>>> as using a relative URL that the problem under discussion arises.
>>
>> What, exactly, would that be relative to?  If your original document
>> is not being accessed as a "file:" URI, then you don't have a base
>> path under that scheme from which to reference relative "file:" URIs.
> 
> Indeed, but it seemed that the OP was addressing the problem of certain 
> file URLs that appeared to be allowed but to have no clear meaning.


file:foo, file:foo/bar file:/foo and file:/foo/bar are *not* allowed 
according to the RFC, but are widely used in practice, and at least 
plausibly useful.


> It 
> seems I am misunderstanding the original problem, but it seemed to be 
> related to file URLs (which presumably had been picked up from other 
> documents, e.g. HTML documents) and which had no obvious Path to anchor 
> them to. So I was trying to think of circumstances where such beasts 
> might be expected to arise, and where a relative URL would not work for 
> some reason. If no such circumstances exist, then the OP's problem does 
> not exist.
> 
I think I identified a case in which these circumstances do exist, it is 
a judgement call whether such a case is in-scope.

Jeremy
Received on Monday, 15 October 2007 10:27:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 13 January 2011 12:15:37 GMT