W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2008

Re: NEW ISSUE: weak validator: definition inconsistent

From: Werner Baumann <werner.baumann@onlinehome.de>
Date: Fri, 11 Jan 2008 21:17:21 +0100
Message-ID: <4787CED1.4010705@onlinehome.de>
CC: ietf-http-wg@w3.org

Lisa Dusseault wrote:
> If we change the definition "Good enough" should be qualified whether 
> the spec means good enough for read-only caching, or good enough for 
> write-without-lost-update.
As I understand the proposal, this would restrict the use of weak etags 
by clients in the same way as the use of all weak validators is 
restricted by RFC 2616:
Clients are only allowed to use them in full body GET requests. I 
believe this is the right thing to do.

> Are there servers that use weak validators besides Apache?
IIS 6.0 uses weak etags, but I have no clear picture about the how. Last 
time I looked at them, it seemed like this: weak etags stay weak until 
IIS is restarted. But I did not really seriously investigate this.

> Should Apache's trick of switching from a weak to a strong validator 
> (with the ability to compare the strong to the weak) be documented?
What I wrote about Apache was based on may understanding, that matching 
weak etags guarantee semantic equivalence. As my understanding changed, 
I would no longer say it this way. Apache uses weak etags because of the 
limited time resolution of 1 second. This would comply with the proposed 
meaning of weak etags, and using a strong etag later would no longer 
look like a "trick" to me. The only remaining issue with Apache's weak 
etags is, they do not even match on a full body GET request, but they 
should. I'm not clear whether this intended or just a simple bug.

Received on Friday, 11 January 2008 20:17:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 11:10:44 UTC