- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Tue, 29 Oct 1996 16:18:38 PST
- To: ben@algroup.co.uk
- CC: gjw@wnetc.com, ejw@kleber.ics.uci.edu, w3c-dist-auth@w3.org
> I like this idea - but it isn't very practical, I fear. If the request is > handled by a CGI, then how does the server know what's been updated? For highly > automated sites, the list of URLs could be huge. a) use patterns to invalidate (mark stale) many URLs at once (this was the '*' in http://host.dom/container/3q96/*) b) Don't invalidate if you don't care about sequential transparency c) If you don't want to invalidate things because there are too many and you don't want to warn about updates, then don't let servers cache them without validation (e.g., send them out originally with max-age=0). So, I disagree that it 'isn't very practical': it's practical in many situations, and when it's not practical, you don't have to use it. Larry
Received on Tuesday, 29 October 1996 20:57:10 UTC