[MITgcm-devel] RE: 2 exp. broken

[Patrick Heimbach]

Dimitris,

I agree it is confusing and has been in the past.
Reasons for combining these were that:
- exf interacted with ctrl
- ctrl interacted with cost
- cal interacted with exf
- autodiff interacted with everthing
- personal histories...

The short answer:
* global_with_exf/code
   provides an example when to use ECCO_CPPOPTIONS.h in a
   cal/exf setting only
* natl_box_adjoint/code
   provides an example with ALLOW_ECCO

The easy part is that in each case all that's needed
is to include ECCO_CPPOPTIONS.h at the end of CPP_OPTIONS.h

The longer answer when my keyboard is fixed,
hopefully this afternoon.

-p.



Dimitris Menemenlis wrote:
> Patrick, I ran into exactly the same problem two days ago in configuring
> global_ocean.cs32x15 to work with pkg/exf surface forcing.
> 
> The confusing an part is that ECCO_CPPOPTIONS.h is associated both with
> a single package, pkg/ecco, as is every other *_OPTIONS.h file and where
> it correctly includes the #ifdef ALLOW_ECCO, "and" with a set of
> packages (pkg/autodiff, pkg/cal, pkg/cost, pkg/ctrl, and pkg/exf) where
> it does not always need #ifdef ALLOW_ECCO.
> 
> pkg/exf and pkg/cal in particular are now used in several forward
> verification experiments that have nothing to do with optimization. 
> Maybe worthwhile to remove the EXF and CAL options from
> ECCO_CPPOPTIONS.h at some point, or would that cause too many
> back-compatibility headaches?
> 
> D.
> 


-- 
Patrick Heimbach ........................... M.I.T
FON: +1/617/253-5259 .......... EAPS, Room 54-1518
FAX: +1/617/253-4464 ..... 77 Massachusetts Avenue
mailto:heimbach at mit.edu ....... Cambridge MA 02139
http://www.mit.edu/~heimbach/ .............. U.S.A