On 14/04/2008, Peter W. Draper <[log in to unmask]> wrote:
> On Mon, 14 Apr 2008, David Berry wrote:
>
> > Renaming a random acsis cube to
> > s20060921_00001_0001(2/3/4/5/6)_850_al, the following works OK for:
> >
> > % maths exp=ia+ib+ic+id+ie+if \
> > ia=s20060921_00001_0001_850_al.MORE.SMURF.EXP_TIME \
> > ib=s20060921_00001_0002_850_al.MORE.SMURF.EXP_TIME \
> > ic=s20060921_00001_0003_850_al.MORE.SMURF.EXP_TIME \
> > id=s20060921_00001_0004_850_al.MORE.SMURF.EXP_TIME \
> > ie=s20060921_00001_0005_850_al.MORE.SMURF.EXP_TIME \
> > if=s20060921_00001_0006_850_al.MORE.SMURF.EXP_TIME \
> > out=fred
> >
> > What command line did you use? Do you know what may have triggered the
> > error message? Is there anything odd about the data files that could
> > have caused the error?
>
>
> David, just a point. Looking at Andy's message:
>
>
> > > !! kappa_mon_2345_1 unexpected failure,
> > > ! MESSYS__BUFOV, message buffer overflow
>
>
> it's clear that this command-line is being passed to a detached monolith.
Good point. I tried it in ICL but the ICL command line editor wont let
me enter such a long command. Allowing maths to prompt for values
rather than supplying them on the command line works fine. If the
problem is to do with message passing, then presumably any app will
have the same problem that uses individual parameters to access each
NDF. We need something that can access the inputs as a single group,
so that the files can be supplied in a text file.
David
|