Mark Martinec wrote:
On Saturday September 15 2007 04:26:59 Daryl C. W. O'Shea wrote:
Mark Martinec wrote:
Right, a few milliseconds during a program startup,
once every few days after a spamd or amavisd restart.
Don't forget that there's a significantly non-zero number of people
running 'spamassassin' for every message on Windows.
17 microseconds of difference! Is it worth discussing it at all?
There are several oportunities for optimization in SpamAssassin,
but this is not one of them.
The penalty for importing things wasn't my point. I was just pointing
out that there are people actually starting the program for *every
single message* rather than the once every few days that you asserted,
so startup time shouldn't be ignored under the guise of it rarely happening.
Daryl