[Thread Prev][Thread Next][Index]

Re: [ferret_users] segmentation fault with fill_xy

Hi Martin,
Thank you for the report. I do see this; it happens only when using the statically-linked executable. We'll check into the cause.


Martin Schmidt wrote:
since some time ferret delivers the function fill_xy, that is a very robust nearest neighbour interpolation or extrapolation filler. It was originally an external function, but is now statically linked. Unfortunately it fails with large data sets. Here an example, artificial, just to demonstrate the effect:

use etopo5
! rose 1 is missing where ocean
let rose1 = if rose gt 0 then rose else 1/0
!show the gaps in the area of the Baltic Sea
shade rose1[x=10:30,y=54:70]

! now fill it again, this requires a mask. Here the mask is 1 everywhere, hence each missing point will be potentially filled
let mask = 0*rose +1
let rose2=fill_xy(rose1, mask, 40) ! repeat the filling procedure 40 times

!and filled
shade rose2[x=10:30,y=54:70]

!and now for a larger area
shade rose2[x=0:50,y=5:70]

/usr/local/bin/ferret: line 13: 25741 Segmentation fault /linsoft/ferret_v640_64_static/bin/ferret_v64 $*

This happens also with previous versions, with suse-linux 9/10 as well as with ubuntu

The dynamically linked version 6.4 works well.

Is there any chance to fix this?

Martin Schmidt

[Thread Prev][Thread Next][Index]

Contact Us
Dept of Commerce / NOAA / OAR / PMEL / TMAP

Privacy Policy | Disclaimer | Accessibility Statement