Bart Schaefer wrote:
There's been some really good stuff in this thread, but I
just can't help throwing in this remark ...
Nothing is actually "bouncing off" the supremum. Procmail
stops scoring when the total score reaches supremum (or
negative supremum), not when any individual condition
changes the score by +/- supremum.
Yes; thanks for restating. I thought that was clear,
but maybe it needed recapitulation. The "bouncing off"
thing is just a phrasal trope. It's meant to tickle the
brain, but not to cause confusion about what's really
happening. It's no more accurate than is, say, "the last
mile" (as "walked" by death-row inmates): we all know
it's not really a mile, nor is the exact distance important.
But the figure of speech _en toto_ carries weight, and that's
what I hoped to create as well with my "bouncing off"
name. You can probably begin, about now, to infer my roots
as a comp lit major rather than a CS one. :-)
The distinction is important because of what happens when
choosing a value for your so-called "oversaturated supremum".
E.g.:
SUPREMUM=2147483647
OVERSUPR=2147490000
:0
* -3000^0
* $ $OVERSUPR^0
* 1^0 this is never tested, supremum was exceeded
{ }
:0
* -4000^0
* $ $OVERSUPR^0
* 1^0 this IS tested, OVERSUPR is still not big enough
{ }
Probably you do understand that, but I wanted to make it
clear to others.
Yes, the distinction is important. I haven't actually been
using an $OVERSUPR in any of these new peregrinations of mine,
but rather, the exact value ($SUPREMUM, above). Perhaps I
should have called it "skipping along the edge of." Hmm.
In any case, thanks, Bart. I apologize if the "bouncing off"
nomenclature bothered you. We all know one can never get there in
any case. Even getting to halfway-to-infinity is still an
insurmountable task. :-) (For any confused here, I'm just kidding
around. Obviously, "infinity" in procmail is not a *real* infinity,
but merely a simulation for your viewing pleasure.)
---
dman (infinitely grateful for such help with explanations)
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail