Re: NCL 5.1.1 seg fault when drawing maps

Hi Adam,

A few people have had trouble with the Cygwin "ncl.exe" binary because
of stack size issues.

I created a new ncl.exe binary that has worked for some folks. If you
want to give it a try, it's on anonymous ftp:

       ftp ftp.ucar.edu
       <log in as "anonymous">
       <Use email address as password>
       cd /pub/scd/haley
       get ncl_new.exe
       quit

Once you have the file, gunzip it, make sure it is executable, and
move it to $NCARG_ROOT/bin:

      gunzip ncl_new_exe.gz
      chmod gou+x ncl_new.exe
      mv ncl_new.exe $NCARG_ROOT/bin/.

Now you can use this new NCL executable instead of the one already
installed on your system:

      ncl_new.exe your_script.ncl

If this works, you can replace your old ncl executable with this new
one. You should first make a copy of your old one:

     mv $NCARG_ROOT/bin/ncl.exe $NCARG_ROOT/bin/ncl_old.exe
     mv $NCARG_ROOT/bin/ncl_new.exe $NCARG_ROOT/bin/ncl.exe

----------------------------------------------------------------------

If this new executable doesn't work, then it would help if I could get
your exact script and data. There's a special resource that you can
set, and it will write out only the data I need to run your script:

   res_at_gsnDebugWriteFileName = "narrbug"

Rerun your script with this resource, and it should create
"narrbug.ncl" and "narrbug.nc". If you can email me these, then I'll
give them a try here.

Thanks,

--Mary

On Mon, 20 Jul 2009, Adam Davis wrote:

> Dear NCL users,
>
> When I run a script that involves creating a map plot using NCL 5.1.1, execution terminates with a segmentation fault at the point where the map is drawn. For instance, when I run the second NARR example script (http://www.ncl.ucar.edu/Applications/Scripts/narr_2.ncl) with the -x option, the script executes normally until the "plot = gsn_csm_contour_map..." line, where a segmentation fault occurs.
>
> The segmentation fault also occurs with version 5.1.1 in several other scripts that I have written that draw maps, which all worked correctly with version 5.0.0 (I don't know about version 5.1.0, though). These scripts use a few different gsn_csm map functions (gsn_csm_contour_map, gsn_csm_vector_scalar_map, and gsn_csm_map), all of which seem to have this issue. I suspect that the problem is occurring when the map is drawn because in a couple of these scripts, I set gsnDraw and gsnFrame to False, then call a map function, add polymarkers and text to the maps, and finally call draw and frame to create the output, and running these scripts with the -x option reveals that the segmentation fault doesn't occur until the first call of draw. I have not encountered any problems drawing XY plots or the like with version 5.1.1 - just maps.
>
> I am running NCL version 5.1.1 on Windows Vista SP1 under Cygwin/X (uname -a outputs CYGWIN_NT-6.0 srbwks45150-01 1.5.25(0.156/4/2) 2008-03-05 19:27 i686 Cygwin), although the same problem also arises on my laptop running version 5.1.1 on Windows XP SP2 under Cygwin/X. No warnings or errors are generated when this issue occurs besides "Segmentation fault (core dumped)".
>
> If you need any additional details about this issue (more example scripts, stack traces, etc.), please let me know and I will be glad to provide them. Thank you in advance for any help you can provide!
>
> Sincerely,
>
> Adam Davis
> BSE Earth System Science and Engineering - Meteorology 2010
> Dept. of Atmospheric Oceanic and Space Sciences
> University of Michigan - Ann Arbor
>
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Jul 21 2009 - 11:37:39 MDT