- From: Schleiff, Marty <marty.schleiff@boeing.com>
- Date: Mon, 14 Aug 2006 15:47:23 -0700
- To: <www-tag@w3.org>
Hi Dave (& All), I only mentioned the stuff about if http ever goes away because section 2.3 of the finding says, "...if for some reason in the future the HTTP protocol becomes unavailable or inappropriate...". I didn't even want to think about that! The finding raises the question about what would happen if the HTTP PROTOCOL goes away, not necessarily the HTTP scheme. My response was about the HTTP PROTOCOL going away, not the HTTP scheme. There is no XRI protocol to go away -- it's protocol independent (sorta - it uses http GETs for resolution, but that could change to some other protocol without impacting the issued XRIs). Although I've read the whole finding, I'm trying to ponder it a little bit at a time, and I'm still way back in section 2. I haven't gotten to the point I'm actually pondering the stuff in section 5 yet, so I haven't yet cogitated the "http redirect versus the xri redirect" that you addressed in section 5. I do suspect I'll find a difference though -- with the http redirect, I'd first go to the old location to find a pointer to the newer location. If the locations changes again, I think a diligent administrator would then re-update the oldest location to point to the newest location, then update the not quite so old location to point to the newest location. If the location changes yet again, would the diligent administrator need to update the 3 prior locations, and so on? I think a difference is that with http, the app looks somewhere for a resource, and then it's exception/error processes would probably be invoked to follow the redirect(s). With XRI resolution the app first pulls back the resource's current XRDS (XML descriptor of the named resource's current service points), and then interacts directly with the current service point. There's only one place (the XRDS) to do the update when an actual location changes. Accessing the resource with XRI is always (ignoring caching) a two-step process, while with http I think it could be a one-to-many-step process. I'll try to be more intelligent about this when I get to section 5. Marty.Schleiff
Received on Monday, 14 August 2006 22:48:20 UTC