[cam-users] GETLATIND: Improperly formatted column string

[Jim McCaa]

Thanks for your response to this, Norm.  Your description is consistent with 
our experience here.  

We have been able to get CAM to run with the ifort compiler on NCAR/CGD 
machines, but only with patched versions of both CAM3.0 and the ifort 
compiler.  The modifications to CAM 3.0 are included in our recent patched 
release ( CAM-3.0.p1 ) available from the main CAM web page.  If you're going 
to pursue this further, I'd recommend using that version of the model.  
Our license for the beta intel compiler has expired, and we've had no success 
with the released compilers (up through 8.1), so we're not currently pursuing 
this.

Jim McCaa


On Tuesday 05 October 2004 8:56 am, Norm Wood wrote:
> Pavel,
>
> On 05 Oct., Pavel Mankevich wrote:
> > Dear All,
> >
> >  GETLATIND: Improperly formatted column string.  Missing
> > underscore character (x
> >  xxE_yyyS)
> >
> >             0
> >  ENDRUN: called without a message string
> > abort:
> >
> > I looked at the source code and found that the reason was
> > in the history.F90 file. It has a subroutine
> > getlatind(..., inname). The argument 'inname' should have
> > an underscore character but in my particular case it is an
> > empty string at all. So could anyone help me understand
> > why it happens? Maybe I set environment incorrectly or
> > something?
>
> I've run into the same situation using the ifort compiler
> with CAM3 and have not been able to resolve the problem (I
> ended up switching back to the pgf90 compiler before digging
> into it too far).
>
> CAM3 introduced some new namelist parameters (fincl1lonlat,
> fincl2lonlat, etc.) for specifying the lat/lon regions for
> which model variables should be written to the history files.
>
> The values for these fincl*lonlat namelist parameters are stored
> in an array fincllonlat in the history module.  At around
> line 1716 of that module, the code tests each string in fincllonlat
> to see if it is empty:
>
> do while (fincllonlat(ff,t) /= ' ')
>    lonlatname(ff) = trim(fincllonlat(ff,t))
>    call getlatind(lonlatname(ff),latind(ff,1),latind(ff,2),latname(ff))
>    call getlonind(lonlatname(ff),lonind(ff,1),lonind(ff,2),lonname(ff))
> ...
>
> For some reason ifort thinks these strings are not empty,
> even if the input namelist left these fincl*lonlat parameters
> undefined.  The result is that getlatind() and getlonind()
> are called when they shouldn't have been.
>
> What is odd is that the code contains other, similar tests on
> namelist parameters fincl, fexcl, fhstpr and fwrtpr in the
> fldlst() subroutine, and no problems occur if these are empty
> strings.
>
> I'm not sure if this problem might be dependent on the ifort version.
> I think I've tried only with version 8.0 under Redhat and Slackware
> linux.
>
> Regards,
>
> Norm
>
> --
> Norm Wood
>
>      Department of Atmospheric Science
>      Colorado State University
>      norm at atmos.colostate.edu
>      phone:  970-491-8480
> _______________________________________________
> cam-users mailing list
> cam-users at ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/cam-users