Re: [ferret_users] apparent bug in frequency_histogram.jnl

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi Keith,
Thank you for the report and the solution. I agree with you about what 
was going on.  This is sure an old script! I've updated our copy and 
we'll include the fix with the next Ferret release.  Writing 
floating-point numbers in a fixed format is always a bit perilous.

Ansley

On 6/15/2011 2:38 PM, Keith Lindsay wrote:
> Hi,
>
> I think I have found a bug in frequency_histogram.jnl. In particular 
> instances, the bars are shifted by two bins horizontally.
>
> The attached CDL file leads to a demonstration of the problem.
>
> The commands
>
> sp ncgen -b hist_test.cdl
> set data hist_test.nc
> go frequency_histogram.jnl var .5 4.5 .1 weight
>
> produce the attached figure. The bins should be centered over 1, 2, 
> and 3, but they are shifted to the right by two bins.
>
> It seems to be problem with missing values in the expression and 
> weight leading to lines in the generated frequency_h.raw file with no 
> space between the value pairs, and then the frequency.awk script 
> generates extra lines of output which ends up shifting the bars.
>
> If I replace the following lines of frequency_histogram.jnl
>
> list/nohead/form=(F10.0,1PG14.7)/file=frequency_h.raw index, wt_index
> list/nohead/form=(F10.0,1PG14.7)/file=frequency_h.raw/append vn+1,0  ! 
> MAX index + 1
>
> with
>
> list/nohead/form=(F10.0,x,1PG14.7)/file=frequency_h.raw index, wt_index
> list/nohead/form=(F10.0,x,1PG14.7)/file=frequency_h.raw/append vn+1,0  
> ! MAX index + 1
>
> I get the expected results.
>
> This does not appear to be a problem with frequency_histogram2.jnl
>
> Keith
>
> ******************************************************************
> Keith Lindsay                http://www.cgd.ucar.edu/oce/klindsay/
> email: klindsay@xxxxxxxx   phone: 303-497-1722   fax: 303-497-1700