|Subject:||RE: [xri-editors] Alternative to Resolution Errors inside Descriptors|
|From:||Wachob, Gabe (gwac...@visa.com)|
|Date:||Jan 31, 2005 9:54:49 am|
Title: RE: [xri-editors] Alternative to Resolution Errors inside Descriptors
I'm not sure how we'd do this - via in RFC2616 specifically requires a list of (protocol version, host:post) - ie 1.1 www.server.com, 1.0 www.proxy.com, etc
Its not clear to me that the syntax there makes any sense since we are talking NOT about HTTP proxies and HTTP versions, but rather about XRI authority servers.
We'd have to specify exactly what gets put in there - and while I suppose we could carry around the domain names of the HTTP servers being used by XRI, that would require specifying some additional behavior on the part of the proxy/lookahead about how to construct the via header.
I think we are too far away from the rfc 2616 concept of a proxy/gateway to meaningfully use (or be required to use) the Via header.
I think this area of debugging, where all the requirements would be "MAY" might be an area where implementations could experiment a bit and then someone (perhaps this TC if its still in existence) could write up a formal spec.
But I think the via header, upon further review, should not be mentioned or used by the spec as of today.
From: Davis, Peter [mailto:pete...@neustar.biz] Sent: Thursday, January 27, 2005 6:11 PM To: Wachob, Gabe; Dave McAlpin; xri-...@lists.oasis-open.org Subject: RE: [xri-editors] Alternative to Resolution Errors inside Descriptors
we should think of this in the context of debugging resolution issues. if via provides insight to (potentially unknown) resolution intermediaries, then yes, i think it is useful to include via.
-----Original Message----- From: Wachob, Gabe [mailto:gwac...@visa.com] Sent: Thu 1/27/2005 7:38 PM To: Dave McAlpin; xri-...@lists.oasis-open.org Subject: RE: [xri-editors] Alternative to Resolution Errors inside Descriptors
I think passing through the error code, as Dave proposes, probably is fine. That's certainly what happens with today's HTTP proxies. I notice the "via:" header for proxy/gateways is REQUIRED (RFC section 14.45, however, and wonder whether we should abide by this requirement. We are sort of a gateway or a proxy, but its not clear that we fall within the intent of gateway and proxy in RFC 2616.