> From: Albert Mietus [mailto:[EMAIL PROTECTED] > Sent: Thursday, December 18, 2003 8:44 AM > > Rainer Gerhards writes: > > > It is regarding the syslog TAG. The format described was > discussed in > > the context of syslog-sign-13 but now also applies to > > syslog-international-00. > > You will understand my vote ;-) > > > So I think we need to make a tradeoff decision: we can either > > a) allow colons in the path name > > xor ;) > > b) allow TAG NOT to be terminated by SP > > I vote A. > > > Selecting > > a) will break compliance for older clients (how many?) > > No it will NOT, > > Syslog-rfc* is (now) step by step, faraway by beeing 100% complice to > old implementations e.g. "long timestamp, DNS hostname, etc. This bit > will not change this.
I fully agree with you as far as the sender's side is concerned. My concern is that we will break compatibility of the receiver receiving existing messages from an old-style syslogd. But, you are right, that old-style syslogd will also break other things. But then, the difference is that we could make proper exclusions/exceptions so that a newer syslogd can successfully parse the message generate by an older system. Now, however, we face the situation that we can not (easily) detect this and the message may be mis-parsed. I am not sure how often "tag:msgmsgsg" actually appears in older messages. I would, however tend to accept this problem. I think it is minor compared to clean design we get be requiring TAG to be terminated by SP, too. So I, too, vote for this. Is there any objection against this? If not, I would like to say this is solved, allowing us to move quickly on with -protocol (as Chris suggested). Rainer
