- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Thu, 17 Oct 96 13:47:37 MDT
- To: Ari Luotonen <luotonen@netscape.com>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
You've made a convincing case that the existing design for Warning can yield bogus Warnings when HTTP/1.0 and HTTP/1.1 caches are combined. (I believe we wrote the HTTP/1.1 caching rules so that an HTTP/1.1 cache in the position of your cache "B" would remove the Warning after doing a successful validation from cache "A", but of course it's too late to apply that to the HTTP/1.0 caches out there.) So please suggest a solution! I can see three options: (1) Live with it. This can only happen when an HTTP/1.0 cache is a client of an HTTP/1.1 cache, and presumably in somewhat unusual cases, so maybe it's a temporary problem. (2) Remove Warning: stale from the protocol, on the grounds that it's better to silently give many users stale pages, instead of bogusly warning a few users about non-stale pages. [I don't consider this option to be a wise choice.] (3) Fix the design so that it works with HTTP/1.0 caches. Perhaps, for example, this means that one can't send a "Warning" to an HTTP/1.0 client (but this would also cause a lack of Warning in cases where it would be apppropriate). One possibility: HTTP/1.1 clients (the only ones that could care about a Warning header anyway) should turn a Reload on a page with a "Warning: stale" into a "Pragma: no-cache". That would cause a few extra cache misses, but would break the infinite loop that you are worried about. -Jeff
Received on Thursday, 17 October 1996 13:58:47 UTC