- From: Adrien de Croy <adrien@qbik.com>
- Date: Tue, 19 Jun 2007 21:58:15 +1200
- To: Adrian Chadd <adrian@creative.net.au>
- CC: HTTP Working Group <ietf-http-wg@w3.org>
Yeah, this all points to some sort of generic config protocol. I think there was a config protocol that Eudora used a while back - it didn't catch on. You'd need something more like SNMP, where configuration points are registered under a global UID schema, or LDAP. Anything that can have a schema and data. DHCP is just so restricted it's hard to see why so much effort is still being poured into it. There are now more options than you can realistically fit into a 576 byte single UDP packet. All to do config. A separate config protocol or framework would be much easier to grow. Use DHCP to discover the config server and go from there. That's basically what AD does, uses DHCP to find the DNS server to do SRV lookups on to find the LDAP server. As long as there is decent client-side API support so that UA developers can hook into it then the solution to auto-config issues is doable. However, to quote Donald Rumsfeld, there are still the things we don't know that we don't know. I'm still open to the possibility that considering intercepting proxies could have some benefits, even if one of those benefits doesn't turn out to be auto-config. That requires contemplation of the issue. Adrien Adrian Chadd wrote: > On Tue, Jun 19, 2007, Adrien de Croy wrote: > >> Actually that proves my point. >> >> this is an example of security problems inherent in low-level protocols >> being solved using high level protocols, e.g. SSL certificates, key >> exchange protocols etc. All of which require the IP config to be >> working, which therefore already required DHCP to be working without >> auth. So, it pretty much makes DHCP auth pointless. >> > > .. and you can push out centralised HTTP proxy server settings via > the same mechanisms. Admittedly its only one platform and I'm not sure > when the AD/Group Policy support popped up, but its certainly doable. > > Me, I'd prefer to see the proxy discovery draft properly worked into > an RFC as there are -plenty- of instances of WPAD being used in the > real world these days. It'd also be nice to have it extensible to proxy > other protocols, such as P2P client proxy discovery (when P2P caches > become all the rage, that is..) > > > > > Adrian > > >
Received on Tuesday, 19 June 2007 09:58:16 UTC