[MITgcm-devel] latest changes in cg2d.F

[Jean-Michel Campin]

Hi,

It looks like my recent changes in cg2d.F (reduce size of overlap
to 1 instead of OLx,OLy) does change the results (truncation differences)
on some platforms (e.g.: beagle) with compiler optmisation on.
Without optimisation, it does not affect the results.
I guess it's not completely unexpected, but just want to make everyone
aware of this.

Completely unrelated: Going to clean-up the old (older than ~3 month ?)
testreport results (from 2006, 2007, 2008 to start with) 
to replace all experiment sub-dir (in each testreport output dir)
with just 1 tar file (it reduces disk-space and number of files/dir
by almost 2 order of magnitude). Any objection ?

Cheers,
Jean-Michel