- From: Graham Klyne <gk@ninebynine.org>
- Date: Sat, 14 May 2005 09:14:12 +0100
- To: John Cowan <jcowan@reutershealth.com>
- Cc: uri@w3.org
A somewhat rambling email follows... but towards the end I note that ftp:
scheme spec proposal draft clearly separates file: and ftp:.
At 01:01 14/05/05 -0400, John Cowan wrote:
>For that matter, on Plan 9 what you see under / depends on what your process
>has attached there. It's not a requirement that "file://localhost/foo" means
>the same thing for everyone even on a single box.
Hmmm... OK.
I *still* think that replacing file: with ftp: is not appropriate in the
general case. Maybe I need to come at this with a use-case argument?
The occasions when I have wanted to use file:// URIs fall into two categories:
(a) to use a browser to examine web pages on a local file system
(b) to develop software than can access local and remote resources in a
uniform way.
(which, now I write them down, I see (a) maybe as a particular case of (b).)
Where the correspondence seems to be particularly important is in
interpretation of the "current working directory" (or local host equivalent
concept) as a URI. Also -- a bit trickier -- in cases where software
attempts to interpret a given string as either a local filename *or* a URI,
depending on the syntax used. (Maybe this latter one isn't a good idea --
better to have distinct specification "modes" for filenames and URIs? --
but I found I needed to do something like this when implementing external
entity handling in an XML parser.)
The common strand here seems to be that the file:// URI needs to have a
globally consistent correspondence with a local user's view of the host
file system. That is, the file:/// URI would be interpreted according to
the implicit context of a local user, and for these purposes would never
contain an explicit authority element.
Hmmm... I suppose you might argue similarly for ftp:/// URIs (without
authority element), in which case I can see grounds for conceding the
"plausibility" you claimed, but I'd need to see that underpinned by
sufficiently rigorous language in both the file: and ftp: URI specifications.
Looking at http://www.ietf.org/internet-drafts/draft-hoffman-ftp-uri-04.txt:
[[
Note that the file: and ftp: URIs are not the same, even when the
target of the ftp: URI is the local host.
]]
which seems to settle the issue
(er, why didn't I look there to begin with ;-).
I can't see anything in the ftp: draft to suggest how ftp:/// should be
interpreted. The only hint I see that it is a legal ftp: URI is in the
broad-brush statement:
[[
An FTP URL follows the standard syntax described in
draft-fielding-uri-rfc2396bis [2396bis].
]]
which implicitly, by omission of any contrary qualification, permits ftp:///.
#g
------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact
Received on Saturday, 14 May 2005 08:20:07 UTC