W3C home > Mailing lists > Public > uri@w3.org > February 2010

Re: fb: URIs? (dispatching on prefix rather than just scheme)

From: Dan Connolly <connolly@w3.org>
Date: Thu, 18 Feb 2010 09:36:01 -0600
To: Sandro Hawke <sandro@w3.org>
Cc: Eran Hammer-Lahav <eran@hueniverse.com>, Graham Klyne <GK-lists@ninebynine.org>, Julian Reschke <julian.reschke@gmx.de>, Ira McDonald <blueroofmusic@gmail.com>, Thomas Fruin <thomasfruin@mac.com>, "uri@w3.org" <uri@w3.org>, David Recordon <davidrecordon@facebook.com>, Dan Brickley <danbri@danbri.org>
Message-ID: <1266507361.7402.139.camel@pav.lan>
On Wed, 2010-02-17 at 12:23 -0500, Sandro Hawke wrote:
> > And put it in the registry no one reads? If you want to stop people
> > from abusing the system, you need to either fix the system or
> > accommodate their use case (i.e. work with them, not against them).
> 
> It looks like the problem is that systems are dispatching on the scheme
> name instead of on the longest leading substring.  If they did the
> latter, then people could use http: or tag: URIs for these apps. 
> 
> Specifically, we hear that
> 
>   http://www.facebook.com/profile.php?id=4
> and
>  fb://profile/4
> 
> are semantically the same [1].  If the systems everywhere could dispatch
> on "http://www.facebook.com/profile.php?id=" as easily as they dispatch
> on "fb:", it seems like the technical side of this problem might go
> away.

Yes... here's hoping we can get from here to there. The road seems
long and looks uphill, though. Sigh.

> (The social part -- people liking to make up URI schemes for personal
> reasons -- would remain.  I doubt that's the real problem, though.)
> 
> Has anyone seen systems which dispatch on leading prefixes?

I think I saw it in a mozilla proxy caching javascript API...

Hmm... it seems that the matching was an arbitrary javascript function.
http://en.wikipedia.org/wiki/Proxy_auto-config

I also have a vague memory that something like this lives (or once
lived) in the Windows Registry.

>   Does anyone
> see a problem with that design?  It would be more confusing in the case
> where no handler is installed, since it could fall back to another
> handler, but I think in general that would be useful.
> 
>     -- Sandro
> 
> [1] http://lists.w3.org/Archives/Public/uri/2010Feb/0018.html
> 
> 


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
gpg D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E
Received on Thursday, 18 February 2010 15:36:04 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:14 UTC