Re: Request to Attend President's Council on Y2K Internet Readiness Roundtable

From: mogul@pa.dec.com
Date: Fri, Jul 09 1999


From: mogul@pa.dec.com
Message-Id: <9907091916.AA01053@youra.pa.dec.com>
To: jg@pa.dec.com (Jim Gettys)
Cc: www-wca@w3.org
Date: Fri, 09 Jul 99 12:16:08 -0700
Subject: Re: Request to Attend President's Council on Y2K Internet Readiness Roundtable 

    I was looking at things from the opposite direction:
    
    If the servers all work correctly and provide 4 digit dates (correctly), 
    we avoid the possible failures in the first place.

Wishful thinking, I believe -- but you're right that it's worth
trying to quantify the scope of the problem.

    More worrying to me actually are proxies that may be buggy, causing
    caching failures, since a large part of the user community is
    behind caching proxies.  A few bugs in a few implementations could
    cause alot of trouble, if the fraction of broken origin server
    traffic is still significant...

Yes, and the potential problems here are not necessarily visible
with passive tracing.  E.g., consider this scenario: all of the
dates in the message headers are 4-digit (therefore showing no
problems in traces - and proxies usually do not themselves add dates
to headers!).  But some misguided coder has decided to parse
a 4-digit date such as
	ABCD
by extracting "CD" and adding 1900.

This kind of problem would only be visible with explicit testing.

-Jeff