[MITgcm-devel] RE: The bad state of "cvs -q update"

[Alistair Adcroft]

Just got off the phone with Chris - he agrees about CPP_EEOPTIONS.h. We are
undecided about SIZE.h - the difficulty is that creating/overwriting files
doesn't work for situations where the files are intentionally in place (e.g.
building in code/ which is a designed feature).

We will discuss the SIZE.h when I'm up tomorrow. This can not be delayed
further - I'm unable to work with CVS at the moment and every delay is just
making it harder to get on with the real tasks of working on the model.

A.
--
Dr Alistair Adcroft            http://www.mit.edu/~adcroft
MIT Climate Modeling Initiative        tel: (617) 253-5938
EAPS 54-1523,  77 Massachusetts Ave,  Cambridge,  MA,  USA

-----Original Message-----
From: Jean-Michel Campin [mailto:jmc at ocean.mit.edu] 
Sent: Wednesday, December 03, 2003 12:02 PM
To: adcroft at mit.edu
Cc: MITgcm-devel at mitgcm.org
Subject: Re: The bad state of "cvs -q update"


Hi Alistair,

I always prefer to have define/undef in .h files rather than 
in the comand line of the genmake/makefile.

In testreport, why not doing a copy of SIZE.h to /build (or /input) 
where thing are compiled, instead of a copy in /code ?
This could be done only when testing MPI, so that we need only 1 
additional copy of SIZE.h meaning SIZE.h_mpi, and can get rid of the 2nd one
.h_nompi. 

By the way, I think this can wait 1 or 2 days, and we can discuss this point
when you are at MIT, this will be quicker and clearer.

See you,

Jean-Michel