[MITgcm-devel] small bugs in pkg/autodiff and pkg/ecco

[Martin Losch]

Hi Patrick,

I think there are small bugs in
autodiff_restore.F
autodiff_store.F
and
ecco/the_main_loop.F
We are trying to create an "ecco-forward simulation", that is with  
these cpp-switches:
#define  ALLOW_ECCO_FORWARD_RUN
#undef  ALLOW_ECCO_DIAGNOSTICS_RUN
#undef  ALLOW_ECCO_OPTIMIZATION

and particularly with this
#undef ALLOW_AUTODIFF_TAMC
because we do not yet have/want to use TAF.
I had to move the position of #ifdef ALLOW_AUTODIFF_TAMC in the two  
autodiff subroutines, in order  to have the routine argument "myThid"  
defined and I moved a #include DYNVARS.h in ecco/the_main_loop.F up  
outside the ALLOW_AUTODIFF_TAMC block to be able to compile.

I am willing to check this in, but I'd rather have you test this or  
even check it in yourself, if you think it  is appropriate. Attached  
is the output of a cvs -q diff -c which hopefully gives you enough  
information for the changes (if not I can send more).

Another observation: the inclusion of PACKAGE_CONFIG.h is a bit  
"unausgegoren" in the various ecco/ctrl/cost-packages. Where's the  
best place to put this statement (we put it into CPP_OPTIONS.h, but I  
am not sure if this is the best place ...)

Martin

-------------- next part --------------
A non-text attachment was scrubbed...
Name: diff.log
Type: application/octet-stream
Size: 2440 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20090826/7098eea4/attachment.obj>
-------------- next part --------------