Re: Question about random number on AS 1.5.5
Re: Question about random number on AS 1.5.5
- Subject: Re: Question about random number on AS 1.5.5
- From: Richard 23 <email@hidden>
- Date: Mon, 26 Feb 2001 06:35:31 -0800
>
No, you're reading the manual correctly. Unfortunately, the improved
>
random number generator in 1.5.5 also changed the seed behavior, so the
>
manual is incorrect. In 1.5.5, if you use "with seed 0", it actually
>
resets the seed to zero. Also, it used to be that if you didn't supply
>
a seed at all, it would use the current time as the initial seed; now it
>
uses zero. I'll have to check and see whether or not this was
>
deliberate.
>
>
In the meantime, your best bet is to use the time as seed. In the
>
startup of your script, say "random number with seed time of (current
>
date)". Then use just "random number" (i.e., with no seed) when you
>
need a random number.
When you guys "improved" random number (I'm assuming you don't view the
seed = zero unless one specifies a seed feature an improvement either),
did you improve the scatter of the result when the magnitude of the
input result is large?
Working on a script which deals with file/folder sizes I substituted a
random number call for the actual Finder size requests to speed up
testing.
Since it included a handler that converts sizes to a string with the
appropriate label (eg, "28 bytes", "2.1 gb", "5.2k" etc) I gave it a
big range:
random number 1.0E+14
But all of the results were between 1E+12 and 1E+14 (in the tb range).
Doesn't seem very random, the scatter I mean.
So this begs the question, just what was improved? 8)
R23