Re: timestamps and timezones (was: time-sync)

[Russ Allbery]

Roger Marquis <[EMAIL PROTECTED]> writes:

 > Also the lack of timeZONE info in the timestamp is a common gripe.  It
 > might be worthwhile to add a short field with this info i.e:
 > 991019_12:54:02_GMT+8.

On the wire at least, it seems to me that timestamps should be in some
universal and easily machine-usable form, such as seconds and fractional
seconds since epoch in UTC.  It might be worth considering something like
TAI too, although getting too far into time issues strays off-focus.

The output format can obviously be whatever is most useful, and I would
imagine that most implementations will offer to write out logs in a format
compatible with existing syslog parsers, but let's not limit the wire
protocol to match the output stream.

(Long term, I'd think people would want to move away from the existing
syslog output format, since even more fatally than lacking the time zone
it lacks the year and requires parsing a symbolic month name.)

-- 
Russ Allbery ([EMAIL PROTECTED])         <URL:http://www.eyrie.org/~eagle/>