Re: Why No new URI protocols [Re: [Usability] Proposal for Spatial Nautilus: Shelf Location]

["Liam R. E. Quin" <liam holoweb net>]

On Mon, 2004-04-12 at 16:22, Alan Horkan wrote:

> Eventually after much searching through the archives I found comment from
> an expert
> http://mail.gnome.org/archives/usability/2003-January/msg00077.html
> Well Nils Pedersen (Sun UI guy) is quoted as saying "URIs are E V I L :)"
> here:
> http://mail.gnome.org/archives/desktop-devel-list/2003-January/msg00277.html
> 
> basically it is exposing a nasty implementation detail that most users
> neither want nor need to know about.  (i expect it was in response to a
> thread that was trying to suggest 'friendly URIs' which rather misses the
> point).  If you are thinking about a new URI for identifying your resource
> you have probably already made your application too complicated.

In particular, a URI is a name for something.  If I can get at Alan's
"shelf" using the same URI that he uses, i.e it's a global name, and
if this is useful, we're onto something.  If it's a misinformed
bletcherous hack like preferences: then don't go near it :-)

The first part (the URI scheme) names the transport protocol (http, ftp,
gopher, telnet, etc), not the semantics.  It might make sense, for
instance, to support gconf://localhost/~user/preferences/ and in the
same way maybe http://localhost:6666/~user/shelf/ or soemthing, but
it's not clear to me that a whole new network protocol is needed.

> "Users should never be expected to type in a gnome URI - or even know
> about them." -- Sander Vesik
Then don't use them.

> URIs dont lend themselves to translation but that goes with the fact that
> they shouldn't be seen anyway.
They are opaque.  The W3C TAG has a note on this.

> this post talks of the navigation difficulties caused by the start-here://
> protocol
start-here also isn't a protocol... I should be able too look it up at
the IETF/IANA registry, right?  if not, *please* gnomies, don't just
bend standards to do what you want without at least reading the specs!!

Microsoft proposed adding some URL schemes early on (1995 or so),
mainly to lock out Netscape at the time.  And there was an
experimental aol: scheme too, of course, with the idea that you
could only follow links to AOL-hosted content if you were on AOL.
It's bad.  Keep the Web open,.

> I'm not sure this is quite relevant to the HIG but can anyone suggest how
> best to publically make clear that many of us think URI's need to be
> avoided?
The problem isn't with URIs themselves -- it's with people making up
things that look vaguely like URIs, calling them URIs, misusing them,
and then saying URIs are bad, as far as I can tell.

Liam

-- 
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
Ankh: irc.sorcery.net www.valinor.sorcery.net irc.gnome.org www.advogato.org