Re: ASN.1 straw poll, take 2

[Kurt Zeilenga <Kurt.Zeilenga@xxxxxxxxx>]

I do not buy into the argument that obsoleting the old ASN.1 would  
imply or necessitate a requirement for implementations to directly use  
the new ASN.1.   As with any (or at least nearly all) use of formal  
language in IETF specifications, the formal language is used only to  
express how data is to be encoded.  Implementations are free to use  
alternative expressions in encoding data so long as what the encodings  
they produce are consistent with encoding that would be produced if  
the definitive expression had been used.

Having two normatively referenced ASN.1 statements leads to confusion  
over which is definitive if they are subsequently found to be  
different.   It would be best to have one and only one referenced  
normatively and hence for this one to be definitive.

I generally favor use of modern ASN.1 as it allows encoding  
requirements to be far more clearly expressed than old ASN.1 and, in  
this case, do find the new ASN.1 module is more clear than the old ASN. 
1 module.

Hence, I am in favor of OBSOLETE.

I note that the LDAP community went through a similar debate when it  
updated LDAP technical specification.  In the end, the LDAP technical  
specification was updated to modern ASN.1, the specification was more  
precise (as allowed by the more modern ASN.1), did not cause any  
implementation to update ASN.1 modules just for update sake, but did  
lead to improvements in interoperability (because the resulting ASN.1  
module was clearer and more precise).

-- Kurt