Re: ui for single variable stored in mulitple files

[Jonathan Callahan <callahan@xxxxxxxxxxxxx>]

Ahh, I see it now.

Sorry Zhangfan.  The problem is in your file names in the descriptor
file.  Although the .xml files used in LAS accept either 'http://...' or
'file:/...', the descriptor file used by Ferret must have a netCDF file
specified for each S_FILENAME.

So your name can be one of three possibilities.

1) the absolute path name 

   /mnt/lasdata0/datasets/adjoint/Have.1997.cdf

2) Just the file (let ferret find it through FER_DATA)

   Have.1997.cdf

3) the DODS URL for this file

Your first stepfile should look like this:

 $STEPFILE_RECORD
   S_FILENAME           = 'Have.1997.cdf',
   S_AUX_SET_NUM        = 0,
   S_START              = 236808,
   S_END                = 245448,
   S_DELTA              = 240,
   S_NUM_OF_FILES       = 1,
   S_REGVARFLAG         = ' ',
 $END 

The magic of DODS is that files whos names begin with 'http://' are
treated by the netCDF libraries as regular netCDF files with data access
handed off to DODS.

Let me know if this doesn't solve your problem.


-- Jon


Zhangfan Xing wrote:
> 
> Hi, Jon,
> 
> Thanks for your reply and I will check through all these points.
> However, here is a quick note about *.des for multiple files:
> LAS does find my *.des file at
> /mnt/lasdata0/datasets/adjoint/adjoint.des
> and just complains about its format:
> 
>  **TMAP ERR: Error in namelist record
>              Error in: FORMAT_RECORD; or file not descriptor file
>              Data set: /mnt/lasdata0/datasets/adjoint/adjoint.des
> set data "/mnt/lasdata0/datasets/adjoint/adjoint"
> 
> It seems the syntax of descriptor file is incorrect, but
> I followed docs exactly. What else could be overlooked?
> Its content is at
> http://eyre.jpl.nasa.gov/download/problem/adjoint.des
> 
> Thanks,
> 
> Zhangfan
> 
> On Mon, 27 Aug 2001, Jonathan Callahan wrote:
> 
> > Zhangfan,
> >
> > There are a couple of changes I'll suggest you make to your xml file.
> >
> > 1) Use xml includes as described here:
> >
> > http://ferret.wrc.noaa.gov/Ferret/LAS/FAQ/xml_includes.html
> >
> > Note that addXml.pl has a '--out_inc' flag to help you make 'includable'
> > xml.
> >
> > 2) Don't use the absolute file names that are generated by addXml.pl.
> > Instead, shorten these so that they include only the file name.  So your
> > xml describing adjoint.des should look like this:
> >
> >   <Have_a units="meters" name="Sea Surface Height"
> > url="file:adjoint#Have">
> >
> > not like this:
> >
> >   **BAD** <Have_a units="meters" name="Sea Surface Height"
> > url="file:/mnt/lasdata0/datasets/adjoint/adjoint#Have"> **BAD**
> >
> > This means that you must make sure your FER_DATA and FER_DESCR
> > environment variables are set properly.  Check
> > lasxml/server/Ferret_config.pl to see these.
> >
> > With the change in 2) I'd expect your adjoint.des file to be
> > accessible.  You'll have to give it a correct grid description of
> > course.
> >
> >
> > As for sister server comparisons.  By default, comparisons are turned
> > off between sister servers because variables may be calculated on the
> > fly and there is no assurance that the tools necessary (scripts,
> > external functions) are available at both sisters.  For variables that
> > are actually found in the files, however, the comparisons can be done.
> > You'll need two things:
> >
> > 1) Your xml for the sistered data must contain DODS address for the
> > data, not local filenames.
> >
> > 2) You'll need to modify lasxml/ui/Template.pm to allow sister
> > comparisons.
> >
> > Here's the modification and some discussion on the subject from the mail
> > archives:
> >
> > http://ferret.wrc.noaa.gov/Ferret/LAS/Mail_Archives/fu_2000/msg00064.html
> >
> >
> > Good luck and let me know if you still have questions/problems.
> >
> >
> > -- Jon
> >