[foaf-protocols] urls as source routing
melvincarvalho at gmail.com
Wed Oct 19 12:04:24 CEST 2011
On 19 October 2011 04:17, Peter Williams <home_pw at msn.com> wrote:
> A long time ago, peter kirstein at ucl-cs taught me the nature of the
> internet (which he worked on at stanford in the late 50s). Together in the
> 90s, we even played with the military bits of the (US) internet, trying to
> figure what bits could usefully apply to the normal life (including crypto).
> At some point, the web intervened (and ldap was born). Out of the ooze
> evolved SSL, linking certs, TP4, ldap and http 1.1. Somewhere in the middle
> web proxies inserted themselves, mostly for spying purposes.
> A few years ago, Henry rammed home to me a particular message, saying (in
> essense): dont use IP address and domain name, Peter. Use URIs, instead.
> This was manna to my ears, as it was a re-incarnation of the PSAP (from the
> OSI tradition). Arguably, the URI syntax was nicer than the PSAP (but not
> much). What matter then as in the URI era was that one used names for a
> sequence of per-layer procedures. Some resolver mapped the name(s) to
> address(s). The brilliance of the web is that this all happens for free (and
> all the usual politics somehow doesnt rear its ugly head).
> The following URI doesnt work (hopefully). BUt it does illustrate, what I
> think I was being taught. Lots of identifiers of what in the internet era we
> would have called "source routing". Its just that the web is doing the
> message/packet switching at layer 8 (rather than some boring router or IPsec
> Now, I believe folks in the linked data object fundamentally to source
> routing. They object to me saying: start a protocol run , that interacts
> with an IDP (a ping pong), that intends to land on a target, that once it
> has a session will deliver a resource named by url. Rather, given one URI,
> the web itself is supposed to orchestrate the routing, and either the foaf
> card or the brower empowered with foaf semantic knowhow will magically
> accomplish what I hand-specified.
> Hmm. May be in unviersity-grade library science, but not in the wild full of
> crappiness and half-baked implementations.
> So, as it stands I like FOAF and source routing (using URIs as
> names/addresses). And, it works (for a million folks). Its even open,and
> flexible. Its even deliverable for costs approaching commodity. But,
> apparently folks in the linked data are yet still miffed. Its still not
> right. This means that even though FOAF and the culture it inculcates works,
> its still not adoptable.
If there's one thing I've learnt in the last few years, it's the the
Web is about tolerance.
Nothing wrong with the ping pong dance but recognize that going direct
has advantages too.
So long as you keep the URI structure intact you're going to have a
Web Scale system.
I think the current tendency is to think about building a system
first, then integration bolted on. The harder way is to think about
integration first, then build your system, with the goal that in the
long term the whole is greater than the sum of the parts, as has been
proved by The Web, hyperlinks and URIs, so far.
> foaf-protocols mailing list
> foaf-protocols at lists.foaf-project.org
More information about the foaf-protocols