If-Modified-Since and forged dated

Hi all,
I am reponding to the longly discussion, and found no individual message
to reply to.
I think, that improving the If-Modified-Since header is a good idea.
Supplying size and checksums (16bit CRC, 32bit CRC, md5 or md3 if anybody thinks
that md5 is too constly) will add some redundancy to protocol.
Generally, redundancy is a good stuff to overcome data integrity problems and
are welcome here, I think.
The question to be answered by WG is: which kind of redundancy shall be built-in
into the protocoll now or later?
The debate on proxies, modifying the Entity-Body is a misunderstandig, I
presume. Proxies, which are application gateways in TCP/IP sense, aren't
gateways in http sense (according to draft), and even if they act as gateways
between 0.9 and 1.x parties, they shouldn't modify (applying conversions, etc)
Entity-Bodies, they will mangle headers only, right?

Dealing with forged dates is an other question.
A picky server implementor may check the timestamp of a file (last modified
timestamp, or mtime in unix terms), and when it looks unreal, can use the
ctime instead if the filesystem supports that. (the ctime can't be forged by users. But users can touch his/her files every day or even more often...)
This can prevent users in creating documents, having bad last-modified dates,
but will not prevent changing system clock, but that can be done only by system
administrators.  I hope, the number of system administrators, willing to do
such misconfiguration is negligible.
I suggest, that this type of workaround should be enabled/diabled in a
configuration option.
Generally, this trouble can be overcome in better education of users, and
elaborating appropriate hit count reporting schemes.
The later is our job, and we shall work on that seriously!
And educating users is the job of sysadmins and webmasters.
We can publich informational RFC-s on this subject, too.
But first, whe shall use Expires: headers as the draft requires whereever
we create web documents! (including the WG's hypertext archives)

Andrew. (Endre Balint Nagy) <bne@bne.ind.eunet.hu>

Received on Tuesday, 15 August 1995 23:39:31 UTC