Paul,
Whoops, I didn't realize that the CRL spec would result in an
ambiguous encoding! Let me try to justify a particular encoding.
The encoding rules in X.208 state that the SEQUENCE OF
construct may be encoded with zero entries, i.e., a zero length
sequence of whatever. So, if we wanted to have that encoding (a zero
length sequence) represent a CRL with no entries, there would have
been no need to mark the SEQUENCE OF CRLEntry as OPTIONAL. Since we
did mark this as OPTIONAL, I argue that a zero length sequence is not
the preferred encoding, although it might be a "legal" encoding
without the larger context presented by the RFC.
In general, if any sort of ASN.1 construct is marked OPTIONAL,
I believe that the right encoding for it the omission of the
construct. If there might be ambiguity in encoded data elements
caused by omission of an OPTIONAL element, then the overall data
structure is, I think, ill formed. So I cannot be enthuiastic about
the encoding approach you adopted, in which a NULL placeholder is
encoded.
Steve