ALbert Mietus | 6 Nov 08:28 2003

new TAG

Hai all,

Rainer made a proposition to rewrite the ABNF for the TAG last week.

I can see the reason for it, and the arguments -- other systems don't use "unix
(dos) slashes" (to make a very short abstract)-- are correct.

However, I would give nthe WG the following in consideration

 * It is only a rewrite. The OLD and the NEW syntax allow both the
   same TAGs, Yes the same strings. Only they breakup in other syntax
 * For "other systems", the differance isn't that big.
  -E.g. the Mac. Native it used the colon to separate a path. However, in
   both the old and the new syntax, that isn't allowed in the TAG (exect
   the specified place). So, the either have to use a short tag (as in
   traditional syslog). Or convert the colon to something else. The already
   do then, when netwerking/cooroparting with Unix/ Windows, ... So that
   isn't complex.
   But remember: the have to do one og both anywow. Both by the old syntax and
   the new one!
  -The same applies to other systems!
 * In both the old and the new syntax, the path-part is a MUST. It is allowed.
  - The new syntax allows it, whil it allows "anything" in the progname part
  - The old syntax suggest it, by using words as "path", path-sep
 * We all algree, the using "the path (partually)" is better, it gives a
   better hint on the programm being run.
  - As the postfix examples show. That exta info is needed. The short hint
    will suffixe to "know" which program it is, not the real sendmail but the
(Continue reading)

Chris Lonvick | 24 Nov 17:06 2003

Separating the syslog Protocol from syslog-sign

Hi Folks,

I've had some discussions with several people and have looked closely at
what we're doing with the definitions of the syslog protocol in the
syslog-sign ID along with what Rainer has been proposing for the
syslog-international ID.

It appears that the idea of the cookies may be used for several future
items.  It doesn't seen appropriate to standardize the cookies in
syslog-sign so we've taken a look at producing separate documents for
syslog-sign and for the definition of the syslog-protocol.

The important part of the Working Group process is that we have consensus
on what we are doing and how we are getting there.  I'd like to ask Rainer
to produce syslog-protocol-00 with the concepts and protocol definitions
that we've been putting into the syslog-sign ID.  If that opens enough
doors for the future extensibility of the protocol then I feel that we can
quickly remove those parts from the current syslog-sign ID to produce a
mechanism that fulfils the intent of authenticating syslog messages and
yet remains transport independent.  I believe that we have consensus on
the signing aspects of syslog-sign at this time and that trimmed ID could
go to Last Call quickly so we can focus our attention on the syslog
protocol.  After all of the discussions we've had I suspect that the
syslog-protocol ID can also move on to Last Call quickly.

Is there anyone opposed to looking at a syslog-protocol-00 ID at this
time?  As said, I think that the effort to produce this should be minimal
since we've had so many good discussions about it in the past, and since
(Continue reading)