http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5356





------- Additional Comments From [EMAIL PROTECTED]  2007-08-24 10:24 -------
(In reply to comment #8)
> There is one detail of a forked environment, which appears
> to need attention.
> 
> A timing report in each child process includes sections that
> were already executed by a parent process long time ago, such as
> 'compile_now'. Also the 'total' time appears to be measured from
> a process start. It seems that some mechanism to reset the timer
> is needed, but I'm not sure what would be the best approach.
> Or perhaps moving it from global to pms ???

I don't think moving it to pms would be a good idea -- there are some
time-consuming actions that occur outside that scope we need to measure (e.g.
copy_config() calls).

I've added a timer_report call to spamd, too, so I can see what you're
talking about... it's pretty gross alright :(

how's about we add a new API that zeroes the timers, which spamd and amavisd can
call?  then before each message scan op starts, we can choose a point to
reset the timers.  sound useful?



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to