RE: FW: Content-MD5 comment (was digest mess)

> ----------
> From: 	Roy T. Fielding[SMTP:fielding@kiwi.ics.uci.edu]
> Sent: 	Monday, January 05, 1998 10:52 AM
> To: 	Jeffrey Mogul
> Cc: 	http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
> Subject: 	Re: FW: Digest mess 
> 
> >One other reason: HTTP/1.1 includes the Content-MD5 header, which
> >specifies the use of base64.  Content-MD5 is optional (and perhaps
> >not even particularly useful?), but its use of base64 is presumably
> >a settled decision.
> 
> Yep, a MIME inheritance.  It turns out that Content-MD5 is not useful
> at all for HTTP/1.1, since the combination of the error-free transport
> layer and length-delimited content is sufficient.
> 
"Error free" and "ones-complement checksum" are not 100% commensurate. Plus,
the existence of proxies menas that the TCP "guarantee", such as it is,
isn't in fact guaranteed anyway.

The MD5 checksum is end-to-end, and much stronger than the transport
checksum.

.