> ok i just performed some more tests and you guys are 
> right, the GET responses during update mode were all 
> 304 except for robots.txt, which is expected i guess. 
> also, total incoming data in update mode was 
> significantly less than mirror mode (3k vs. 42k).

This is generally what happends (even if ALL **LOCAL** 
files are being 'quickly' rescanned)

> the site where i saw almost identical download times 
> for mirror vs. update modes was 
> <http://www.oldskool.org/pc/help/oldonnew/>, which as it 
> turned out consisted mostly of .shtml files.

Right - the server does not 'understand' update requests, 
and always send 'fresh' data. Even the httrack's 'Update 
hack' can not avoid that, because the server doesn't even 
give any content-length field:

HTTP/1.1 200 OK
Date: Thu, 16 May 2002 20:11:43 GMT
Server: Apache/1.3.20 (Unix) DAV/1.0.2 mod_ssl/2.8.4 
OpenSSL/0.9.6 PHP/4.0.6 mod_perl/1.25
Connection: close
Content-Type: text/html

Update problems (that is, data being retransfered) is 
generally due to bad server implementations. This is a 
shame, as Etag is potentially the perfect-and-flawless 
solution for updating/caching (even for POSTed data, data 
with cookies, user parameters..)