Bart Schaefer wrote:
I got curious and tried this with my procmail 3.22, RH Linux 6.2.
With the same input every time:
....
* .1^1 xx Score: 9 9 "xx"
* .11^1 xx Score: 10 10 "xx"
* .2^1 xx Score: 19 19 "xx"
These are at least close and consistent. Did you try any other
bases that I have really wacky problems with? (eg. 0.55)
* 1^.9 xx Score: 1 1 "xx"
This is really out of wack for you too since it should be '10'.
* 10^.9 xx Score: 91 91 "xx"
Ditto here, should be '100'.
* 100^0.09 xx Score: 109 109 "xx"
* 100^0.009 xx Score: 100 100 "xx"
The first should be 110 with rounding but they are close, and
again the same as mine.
I'm only guessing, but I'd suspect a rounding error when storing the
intermediate value between matches. Different floating point precision
could explain why Daryle's results vary so much more wildly than mine,
and why it differs for different exponent values.
Well our exponent results are the same (and incorrect). The fractional
base effect does seem highly unstable from setup to setup though. I
would be interested to know if you have trouble with any of the 100
values from .01 to .99.
Here is my output for fractional bases of .01 to .99 and an exponent
of 1 (eg. .06^1) and the fractional exponent of .9 with bases from
1 to 10 (eg. 3^.9) all with 100 matches:
procmail: Score: 0 (should be 1)
procmail: Score: 1 (should be 2, etc.)
procmail: Score: 2
procmail: Score: 4
procmail: Score: 4
procmail: Score: 5
procmail: Score: 0
procmail: Score: 8
procmail: Score: 8
procmail: Score: 10
procmail: Score: 0
procmail: Score: 11
procmail: Score: 13
procmail: Score: 0
procmail: Score: 15
procmail: Score: 15
procmail: Score: 17
procmail: Score: 18
procmail: Score: 0
procmail: Score: 19
procmail: Score: 0
procmail: Score: 0
procmail: Score: 0
procmail: Score: 24
procmail: Score: 25
procmail: Score: 25
procmail: Score: 0
procmail: Score: 0
procmail: Score: 28
procmail: Score: 30
procmail: Score: 31
procmail: Score: 31
procmail: Score: 32
procmail: Score: 33
procmail: Score: 0
procmail: Score: 36
procmail: Score: 0
procmail: Score: 0
procmail: Score: 38
procmail: Score: 39
procmail: Score: 0
procmail: Score: 0
procmail: Score: 0
procmail: Score: 0
procmail: Score: 45
procmail: Score: 0
procmail: Score: 47
procmail: Score: 48
procmail: Score: 0
procmail: Score: 50
procmail: Score: 0
procmail: Score: 51
procmail: Score: 52
procmail: Score: 0
procmail: Score: 0
procmail: Score: 0
procmail: Score: 57
procmail: Score: 58
procmail: Score: 59
procmail: Score: 59
procmail: Score: 60
procmail: Score: 61
procmail: Score: 63
procmail: Score: 64
procmail: Score: 65
procmail: Score: 66
procmail: Score: 67
procmail: Score: 68
procmail: Score: 0
procmail: Score: 0
procmail: Score: 71
procmail: Score: 72
procmail: Score: 0
procmail: Score: 0
procmail: Score: 75
procmail: Score: 0
procmail: Score: 0
procmail: Score: 77
procmail: Score: 78
procmail: Score: 79
procmail: Score: 80
procmail: Score: 0
procmail: Score: 82
procmail: Score: 0
procmail: Score: 0
procmail: Score: 0
procmail: Score: 86
procmail: Score: 0
procmail: Score: 88
procmail: Score: 90
procmail: Score: 90
procmail: Score: 0
procmail: Score: 92
procmail: Score: 94
procmail: Score: 94
procmail: Score: 96
procmail: Score: 96
procmail: Score: 0
procmail: Score: 0
procmail: Score: 100
----------------
procmail: Score: 1 (should be 10)
procmail: Score: 11 (should be 20, etc.)
procmail: Score: 21
procmail: Score: 31
procmail: Score: 41
procmail: Score: 51
procmail: Score: 61
procmail: Score: 71
procmail: Score: 81
procmail: Score: 91
--
Daryle A. Tilroe
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail