W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2010

Re: proposal for issue #178

From: David Morris <dwm@xpasc.com>
Date: Mon, 7 Jun 2010 09:10:05 -0700 (PDT)
To: "'HTTP Working Group'" <ietf-http-wg@w3.org>
Message-ID: <Pine.LNX.4.64.1006070906260.743@egate.xpasc.com>

Etag proves nothing if the servers are out of sync data wise but the 
Etags agree for what ever reason. Which I consider much higher probability
than corruption in a single segment response. An MD5 or other 'checksum' 
of similar or better strength when applied to the reconstructed entity 
will 'prove' it to be completely received and reconstructed correctly.

Etag only indicates that the servers believe they are serving the same
binary bits, not that they did, or the the reconstruction was bug free,
etc.

On Mon, 7 Jun 2010, Julian Reschke wrote:

> On 07.06.2010 17:05, David Morris wrote:
> > ...
> > Fact of the matter is that the probablitly of getting one chunk in error
> > is very low in comparison with the origin servers being out of sync being
> > the reason for the error. Without an MD5 for the whole entity there is no
> > was to check the complete entity for correctness.
> > ...
> 
> Hmmm. Etag?
> 
Received on Monday, 7 June 2010 16:10:40 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:20 GMT