[EMAIL PROTECTED] schrieb am 04.11.02 08:18:09: > Message: 5 > Date: Sun, 3 Nov 2002 13:12:11 -0800 > From: Guy Harris <[EMAIL PROTECTED]> > To: Ronnie Sahlberg <[EMAIL PROTECTED]> > Cc: Jean-Baptiste Marchand <[EMAIL PROTECTED]>, > [EMAIL PROTECTED] > Subject: Re: [Ethereal-dev] Patch for uuid strings in dcerpc dissectors > > ...although, if a UUID corresponds to a particular protocol, no matter > what version - just as the program number in ONC RPC corresponds to a > particular protocol, no matter what version - we could split the DCE > registration functions into a "dcerpc_init_uuid()" that takes a protocol > number, and ett value, and a UUID, and registered the UUID for that > protocol, and another routine that takes a UUID, a version number, and a > "dcerpc_sub_dissector *" and registers the dissectors for that version > of the protocol. (I don't know which one would take the "opnum_hf" > argument; if different versions of the protocol can have different names > for particular operation numbers, it'd be the second routine, along the > lines of what's one for ONC RPC, otherwise it'd be the first routine.) > Please keep in mind, that UUIDs itself are not necessarily corresponding to a protocol at all.
For example: The object id in the DCE-RPC header is also a UUID, but is of course no protocol and has no version. DCOM uses the UUID's in other places as well: IID's, CLSID's, ... Especially the CLSID's have names which could be registered with the same mechanism. Maybe it is a better idea, to register only a UUID as a kind of anchor, with a name and a list of subdissectors and nothing more, and the subdissector (or in terms of DCOM, an interface), which holds everything else. Regards, ULFL ________________________________________________________________ Keine verlorenen Lotto-Quittungen, keine vergessenen Gewinne mehr! Beim WEB.DE Lottoservice: http://tippen2.web.de/?x=13
