- From: Daniel W. Connolly <connolly@beach.w3.org>
- Date: Thu, 18 May 1995 17:44:44 -0400
- To: Chuck Shotton <cshotton@biap.com>
- Cc: Ari Luotonen <luotonen@netscape.com>, brian@organic.com, dwm@shell.portal.com, john@math.nwu.edu, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
In message <v0211011eabe16b310d8c@[129.106.201.2]>, Chuck Shotton writes: > >So, how do you deal with multi-fork files, byte ranges from generated >output streams, ranges that are interpreted as records in a database or any >other number of problems? You don't. Or at least: you don't have to. The proposal, as I see it, is: _If_ you're going to serve up ranges of bytes, please write youre URLs this way: ... . If you're not going to serve up byte ranges, then you don't need to bother with this proposal. Your clients/servers/proxies will continue to operate reliably. There's one more optimization: if you're a proxy, and you've looking at a byte range request xxx;byterange=S-E and you've got up-to-date cached responses with URLs of the form xxx;byterange=Si-Ei and the intervals [Si, Ei] cover the interval [S,E], you can compute the response to the current request from your cached requests. This optimization is only correct if nobody uses URLs of the form xxx;byterange=X-Y except for this purpose. And that's the only reason they needed to tell the rest of us about this at all. Otherwise, it could have been a completely server-private feature. Is that in line with what you meant, Ari? Dan
Received on Thursday, 18 May 1995 14:49:13 UTC