On Thursday 12 April 2001 10:48, Guy Harris wrote:
> On Wed, Apr 11, 2001 at 08:10:02PM -0700, Doug wrote:
> > I guess the question is how stateful are the protocol modules in
> > Ethereal,
>
> Some of them are quite stateful.
>
> However, there are many protocols that aren't, and a dissector-generator
> might work for them; even though not *all* protocols could be handled by
> having a "generic" dissector read an XML description, many of them
> perhaps could be.
I don't want to spoil the hype about XML, but there is a different approach
to formal description of protocols, named ASN.1. At the moment, ASN.1 is not
very useful for analysing protocols because the encoding of protocols
described by ASN.1 is fixed (one of PER, BER, DER -encoding). That will be
addressed by EDM (encoding definition modules) and ECN (encoding control
notation), ITU-T Rec. X.692, ISO/IEC 8825-3. There aren't many protocols by
now which are described with ASN.1 (UMTS, GPRS, Bluetooth, BACnet...).
Hopefully, ASN.1 is going to be an important standardisation language for
future protocols. That would help a lot for implementation of protocol stacks
and Ethereal dissectors because of automatically generated source code
directly from the protocol standard. I think that the snmp-dissector is
generated that way, isn't it?
Cheers, Hartmut Mueller
PS: For information about ASN.1 please have a look at
http://asn1.elibel.tm.fr/en/book/ from Olivier Dubuisson. There is also a
mailing list at asn1@xxxxxxx.