Hi All :-) first of all, I wish I apologize for not giving any feedback to the list since many weeks. I was not very fine, and so I was not able to manage mails as I'd rather prefer. Inevitably, protocols present some assumptions. Sometimes it's possible to leave some of them out of consideration. In this e-mail and in the next one I'll try to "address" a couple of them that affects also syslog. -Network Protocol Assumption- The widespread success of the IP Internet has put it in the position of being one of the most important network protocols used by application protocols, such as the traditional syslog based on UDP. Darren New and Marshall T. Rose also worked on describing two mappings of the syslog protocol to TCP connections. Anyway both these transport protocols do not rely necessarily on IP. For example RFC 1791 and RFC 2147 describe respectively TCP and UDP over IPX networks with fixed path MTU and over IPv6 Jumbograms. However, all the current drafts we are working on suffer of this assumption. They in fact inherit from the traditional syslog the same syntax for the MSG part of syslog packet, where network address are eventually specified in the HOSTNAME field (at present, in our drafts network address are always IP address). Getting rid of this assumption can make syslog protocol more portable and flexible on heterogeneous networks. What about replacing all instances of 'IP address' with 'network protocol address' in new version of syslog drafts? We could also add examples of syslog packet containing several kind of network address in the MSG part; and add a new section to address this assumption. Sincerely, alfonso ~ ~ ~ ~
