[MITgcm-devel] thsice AD test

[Jean-Michel Campin]

Hi Patrick,

There are still some problems with global_ocean.cs32x15/input_ad.thsice:
a) it does not run with MPI (something wrong happen in thsice in some
  recomputations, and then the surface temp does not converge normally).
b) does not look "solid": the little changes I made last Friday
  in thsice_solve4temp.F (changing the debug/error print messages)
  results in different results on faulks with g77 & dickens with gfortran
  so that it does not "pass" anymore (only 9 digits for the gradients).
When you have time, if you could have a look at this.

Cheers,
Jean-Michel