bc is an arbitrary precision "calculator", so floats are not represented in it as in the other "normal" programs. You should not look at is as a reference.
In fact you can make any precision calculation you want, just try (inside bc): scale=100 1/3 This will give you 1/3 with 100 digits!! Cheers, Xavier On 5/3/07, Adam Ginsburg <[EMAIL PROTECTED]> wrote:
I have a strange problem that seems not to be isolated to PDL, but causes problems when using PGPLOT. I'm making a postscript plot, but when I try to plot certain values (e.g. -.360000000000001), it makes a bad plot - specifically, it makes an irrational bounding box (44 -154618798 154619554327). I've found out that the reason I have such a weird value is that apparently linux (kernel vers. 2.6.17-11-386) does something weird with some values. Example: 9.37-9.73 = -.360000000000001 instead of -.36 1.1-1.2=-0.0999999999999 1.10-1.20 = same 1.10-1.21 = -.11 If I float() the numbers, the precision is worse. The above values happen in PDL and OpenOffice, but not in bc (bash calculator), so presumably this has something to do with the way doubles/floats are treated in these programs, and bc doesn't use the same method. Can anyone explain to me why that happens or how to fix it? Or, failing that, some workaround to get PGPLOT not to put out bad bounding boxes? Is my problem unique? Thanks, Adam _______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
