[ng-spice-bugs] Bug in polyfit.c (fwd)
Oops,
I think I did not noticed this message.
---------- Forwarded message ----------
Date: Sat, 30 Sep 2000 01:03:37 +0200
From: Daniele Gordini <dgordini@guest.deis.unibo.it>
Reply-To: ng-spice-bugs@ieee.ing.uniroma1.it
To: ng-spice-bugs@ieee.ing.uniroma1.it
Subject: [ng-spice-bugs] Bug in polyfit.c
Hello!
I've found a little bug in ng-spice-rework-12.
Using the Fourier Analysis all the magnitudes of the
harmonic components are calculated as zero, so most of
the results show up as a meaningless NAN.
I've traced the bug to line 20 of the utility file
"src/maths/poly/polyfit.c", which reads:
memcpy((char *) ydata, (char *) mat2, n * sizeof
(double));
which is incorrect, it should be:
memcpy((char *) mat2, (char *) ydata, n * sizeof (double));
i.e. source and destination of the "memcpy" are swapped.
This causes the zeroing of the input data before the
data has a chance of being used.
Thank you for your job!
Best regards,
D. Gordini
Partial thread listing: