Re: netCDF-3 ncdump bug

> To: Russ Rew <address@hidden>
> Subject: Re: netCDF-3 ncdump bug
> From: Mark Borges <address@hidden>

Hi Mark,

> >> Russ Rew(RR) wrote:
> RR> Hi,
> RR> There's a bug in ncdump in the netCDF-3.3 release.  If a variable has
> RR> more than 1000 values in its last dimension, ncdump will incorrectly
> RR> output a ";" instead of a "," separator after one of the values.
> 
> RR> A description of the problem and a patch to fix it are available from
> RR> the "Known Problems with the netCDF 3.3 Distribution" Web page at
> 
> RR>     http://www.unidata.ucar.edu/packages/netcdf/known_problems.html
> 
> RR> This is where we'll be accumulating any other problems and workarounds
> RR> reported with the new release.  A subsequent minor release will include
> RR> the accumulated fixes.
> 
> RR> Thanks to Alain Coat for reporting the bug.
> 
> Hi Russ,
> 
> I tried to apply the patches to netcdf-3.3 (the one with this md5 stamp):
> 
>   $ md5 ~/misc/src/netcdf-3.3.tar.Z
>   MD5 (/home/mdb/misc/src/netcdf-3.3.tar.Z) = 9449e51fc4dcbf56da85892ac471ff1c
> 
> but it failed for nearly all the hunks:
> 
>   $ patch --verbose -b  < ../../netcdf-3-patches
>   Hmm...  Looks like a normal diff to me...
  ...
  
> (I'm using GNU patch-2.2). Would it be possible for you to post a
> context or unified diff instead? That way if it fails I could manually
> apply the patch with some confidence. Alternatively, if the next minor
> release is just around the corner (within a week or so) I could just
> wait for that.

Strange, MD5 gives the same signature for me, and the vardata.c patch
worked fine for me using /bin/patch from a SunOS 5.5 platform:

 $ md5 ~ftp/pub/netcdf/netcdf-3.3.tar.Z
 MD5 (/home/ftp/pub/netcdf/netcdf-3.3.tar.Z) = 9449e51fc4dcbf56da85892ac471ff1c
  
 $ patch < ~/tmp/patch1
   Looks like a normal diff.
 done

But I've changed the patches on the "Known Problems" web page to be
context diffs instead:

   http://www.unidata.ucar.edu/packages/netcdf/known_problems.html

which should make it easier for any version of patch.

By the way, I'm not familiar with the terminology "unified diff".
I'm guessing it refers to diffs suitable as input to the "ed" editor.

The next minor release will probably come out next week.  It will have
bug fixes to the ncgen "-f" and "-c" options, some ports to a few
additional platforms, miscellaneous fixes, and a new RELEASE_NOTES
document.

Thanks for reporting the problem!

--Russ

_____________________________________________________________________

Russ Rew                                         UCAR Unidata Program
address@hidden                     http://www.unidata.ucar.edu