I agree with Andrew,

We so often have seen those limits that we do not envision to become
limits turn into hard ones... Did you ever envision having 1 GB of
memory in your laptop...

Rainer

> >Maximum TIMESTAMP field length of 30 characters (that should give 4
> >secfrac characters if the rest of the field is maximally filled -
> >having 5 characters in the offset with a "+"/"-" rather than
> "Z".) [I
> >just like bounding these things. :-) ]
>
> I know that 30 is a nice round number and all, but if we are
> going to limit the length, we should make it 32 chrs. This
> gives 6 digits for the secfrac. Millisecond timing is fine
> now, but with 3GHz+ CPUs and 10Gb Ethernet around these days,
> I think microsecond timing may be useful. The PIX blade on a
> Cat6000 for example can *theoretically* log over 30K messages
> per second with debug on. Greater secfrac granularity may be
> required sooner than we think.
>
> Cheers
>
> Andrew
>
>
>


Reply via email to