On 5/11/04 12:47 PM, "Robert Sayre" <[EMAIL PROTECTED]> wrote:
>>> That's not what the Pace says. The Pace says you can change the ErrorURI >>> for a given feed whenever you want. I would probably change the value of >>> the of the ErrorURI every time the feed was generated. >> >> >> still doesn't help, unless you want to eliminate proxy caches. > > False. sorry, I must have misunderstood you -- I thought you said you would change the errorURI to prevent two or more clients receiving the same errorURI, thus avoiding the problem of having to respond to those reports (since a given client shouldn't report to the same URI more than once). (If that is not the reason for you changing the errorURI each time, then just what is your reason for doing so?) All this based on your assertion that the pace doesn't say what I claimed... >>> the Pace describes use of a 410 GONE response once the error had been fixed >>> > That's not what the Pace says. Which is odd, since the pace says this: > 5.6.4.3 410 > > The server is no longer accepting error reports to this URI, and will not > resume accepting reports in the future. The client SHOULD NOT send any further > requests to this URI. and then follows that with an extended example illustrating just that. It might also pay to backtrace on this thread, which started with you stating: > I don't follow your logic. I would just ignore hits from referrers I > don't care about. which then prompted my very sensible question: >>> ... is it proper http behaviour to vary the response (200 OK vs 410 GONE) >>> for a given specific URI according to the Referer header value? (I'm still interested in the generic HTTP answer to that question, btw) So Robert, say your server receives a GRUMBLE to your ErrorURI, and say you have already fixed the problem, and say the referer header is some third-party URI ... just what response will you provide to the request? > Is your problem that the server is still receiving ERRORs after the > problem has been fixed? That's just an accurate reflection of reality. That's not my question (nor my problem). My question was is it http-proper to vary the response depending on the referer header. What's the answer to that question, and this time lets not get distracted by red herrings like changing the ErrorURI. e.
