RE: [ncl-talk] Buggy warning still present

Mary,

Sure, no worries. Simply reference the following two people...

Daniel Leuenberger and Oliver Fuhrer, MeteoSwiss

I've attached the NCL script we used to generate the plot. Note that it
builds upon a library which is automatically loaded upon startup of NCL.
Hope to see you soon sometime here in Europe for an NCL training
course!!! ;-)

Oli

________________________________________

Oliver Fuhrer
Numerical Models

Federal Departement of Home Affairs FDHA
Federal Office of Meteorology and Climatology MeteoSwiss

Kraehbuehlstrasse 58, P.O. Box 514, CH-8044 Zurich, Switzerland

Tel. +41 44 256 93 59
Fax +41 44 256 92 78
oliver.fuhrer@meteoswiss.ch
www.meteoswiss.ch - First-hand information
  
 

> -----Original Message-----
> From: Mary Haley [mailto:haley@ucar.edu]
> Sent: Freitag, 8. Mai 2009 16:22
> To: Fuhrer Oliver
> Subject: Re: Buggy warning still present
>
> Hi Oliver,
>
> Thanks so much for sending this! Would it be okay if I
> include it in
> my suite of plots created by users, that I occasionally
> show to people during my talks? I will be happy to attribute you in
> any way you like.
>
> We are happy to see the activity in ncl-talk, even if it is more bug
> reports. We want to clean up the code as much as possible.
>
> I will share this plot with my co-workers, and thanks for your
> positive comments!
>
> --Mary
>
>
> On May 5, 2009, at 9:54 AM, <Oliver.Fuhrer@meteoswiss.ch>
> <Oliver.Fuhrer@meteoswiss.ch> wrote:
>
> > Dear Mary,
> >
> > Thanks for your two replies! The ncl-talk mailing list seems to
> > give the
> > impression that there are only problems with NCL, well here's some
> > good
> > news for a change... Today our management decided to use NCL for
> > operations and start producing some of the plots using NCL. The
> > attached
> > PNG is an example of one of the first products which is produced
> > hourly.
> > It compares calibrated radar observations with an analysis and 8
> > forecasts of different initial times of our operational model
> > running at
> > a horizontal resolution of 2km. The black line (which is
> somewhat pig
> > shaped) is the outline of Switzerland. We are very happy with the
> > result
> > and are hoping to introduce new plots rather soon! So a big
> thanks to
> > all of you folks helping us make cool plots with NCL!!!
> >
> > And thanks also for the hint concerning the local declaration!
> >
> > Cheers,
> > Oli
> >
> > ________________________________________
> >
> > Oliver Fuhrer
> > Numerical Models
> >
> > Federal Departement of Home Affairs FDHA
> > Federal Office of Meteorology and Climatology MeteoSwiss
> >
> > Kraehbuehlstrasse 58, P.O. Box 514, CH-8044 Zurich, Switzerland
> >
> > Tel. +41 44 256 93 59
> > Fax +41 44 256 92 78
> > oliver.fuhrer@meteoswiss.ch
> > www.meteoswiss.ch - First-hand information
> >
> >
> >
> >> -----Original Message-----
> >> From: Mary Haley [mailto:haley@ucar.edu]
> >> Sent: Dienstag, 5. Mai 2009 16:56
> >> To: Fuhrer Oliver
> >> Cc: ncl-talk@ucar.edu; deMorsier Guy
> >> Subject: Re: Buggy warning still present
> >>
> >>
> >> Hi Oliver,
> >>
> >> Thanks for providing a nice short script to illustrate the problem.
> >> It does the same thing on my system.
> >>
> >> One possible word of caution: I don't think you want to declare
> >> variables that are passed into a subroutine as "local".
> I'm not sure
> >> it is clearly defined what this means, but I thought that variables
> >> declared local were "cleaned up" before leaving the
> >> subroutine. Removing these "local var"s from your script doesn't
> >> change anything, so we still need to deal with the problem at hand.
> >>
> >> We are actually planning a 5.1.1 release in about a month, so
> >> maybe we can get this fixed by then.
> >>
> >> --Mary
> >>
> >>
> >> On Tue, 5 May 2009 Oliver.Fuhrer@meteoswiss.ch wrote:
> >>
> >>> Hi ncl-talk,
> >>>
> >>> In 5.0 there was a erroneous warning when the dimension name of a
> >>> variable was changed in a subroutine and returned to the
> >> caller. As NCL
> >>> passes variables by reference, there should be no warning.
> >> This warning
> >>> has been removed in 5.1. But obviously it is still present, if a
> >>> subroutine calls a subroutine. The warning is the following...
> >>>
> >>> warning:VarVarWrite: Dimension names for dimension number
> (0) don't
> >>> match, assigning name of rhs dimension to lhs, use
> "(/../)" if this
> >>> change is not desired
> >>>
> >>> The example below illustrates the problem. The problem
> only arises,
> >>> since the dimension name is changed in sub2 called from
> sub1. If the
> >>> call of sub1 on the main script level is replaced with
> >> sub2, no warning
> >>> is issued.
> >>>
> >>> This might seem a detail, but as we have a library build
> around the
> >>> COSMO model which changes quite frequently the names of
> >> dimensions, we
> >>> have warnings issued all over the place which clutters the
> >> useful output
> >>> considerably.
> >>>
> >>> Kind regards,
> >>> Oli
> >>>
> >>> ------------ dimname_problem.ncl ---------------
> >>>
> >>> undef("sub2")
> >>> procedure sub2( var:numeric )
> >>> local var, s
> >>> begin
> >>> delete(var!0)
> >>> var!0 = "two"
> >>> end
> >>>
> >>> undef("sub1")
> >>> procedure sub1( var:numeric )
> >>> local var
> >>> begin
> >>> sub2(var)
> >>> end
> >>>
> >>> a = (/1,2,3,4/)
> >>> a!0 = "one"
> >>> sub1(a)
> >>>
> >>>
> >>> ________________________________________
> >>>
> >>> Oliver Fuhrer
> >>> Numerical Models
> >>>
> >>> Federal Departement of Home Affairs FDHA
> >>> Federal Office of Meteorology and Climatology MeteoSwiss
> >>>
> >>> Kraehbuehlstrasse 58, P.O. Box 514, CH-8044 Zurich, Switzerland
> >>>
> >>> Tel. +41 44 256 93 59
> >>> Fax +41 44 256 92 78
> >>> oliver.fuhrer@meteoswiss.ch
> >>> www.meteoswiss.ch - First-hand information
> >>>
> >>> _______________________________________________
> >>> ncl-talk mailing list
> >>> List instructions, subscriber options, unsubscribe:
> >>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >>>
> >>
> >> <cosmorad.png>
>
>