From: hallam@w3.org
Date: Sun, 3 Sep 95 12:52:12 PDT
To: cypherpunks@toad.com
Subject: Re: ASN.1 and Kerberos version 5
In-Reply-To: <ac6e483800021004fa65@[18.162.1.1]>
Message-ID: <9509031951.AA23373@zorch.w3.org>
MIME-Version: 1.0
Content-Type: text/plain



I don't think that the concept of ASN.1 is as bad as Jeff makes out. If it worked 
then ASN.1 would be very very usefull. But is just plain don't.

ASN.1 is worse than useless, it means that a very good idea is rendered unusable 
because of a baddly botched implementation.

The ambiguities of the ASN.1 spec are at least as bad as Jeff makes out. I have 
attempted to implement an ASN.1 compiler but I have little cofidence in its 
correctness because the structure of ASN.1 is so unweildy. It is not just ANY 
that causes problems, IMPLICIT is a complete cock up.


ASN.1 is poor because it is unecessarily complex, has little intelectual 
coherence and has been extended in a manner which conflicts with the original 
design principle. 

Is it any coincidence that ASN.1 backwards is the name of a well known 
organisation? Also the only person who has defended ASN.1 to my face happened to 
work for that organisation once.


So the motto is: ASN.1 - Just say NO!

		Phill