Merge lp:~wence/fluidity/fix-momcg-les-no-dynamic into lp:fluidity

[Lawrence Mitchell]

This should fix tides_in_the_Mediterranean_Sea example on ifort.

[Jonathan Bull (j-bull08)]

Good spot. Why does this break the tides test?

[Lawrence Mitchell (wence)]

I think the breakage was because previously we were not running the tides test as a longtest with intel fortran. It's possible that gfortran is more lenient about unassociated pointers that are not read.

[Lawrence Mitchell (wence)]

On 1 Aug 2012, at 11:27, Lawrence Mitchell wrote:

> I think the breakage was because previously we were not running the
> tides test as a longtest with intel fortran. It's possible that
> gfortran is more lenient about unassociated pointers that are not
> read.

Buildbot green http://buildbot-ocean.ese.ic.ac.uk:8080/builders/fix-
momcg-les-no-dynamic/builds/0

[Jonathan Bull (j-bull08)]

I understand wence's arguments for this and am happy to approve it.

review: Approve

=== modified file 'assemble/Momentum_CG.F90'
--- assemble/Momentum_CG.F90	2012-05-23 14:01:16 +0000
+++ assemble/Momentum_CG.F90	2012-07-25 15:08:26 +0000
@@ -439,6 +439,9 @@
            call leonard_tensor(nu, x, tnu, leonard, alpha, les_option_path)
 
            ewrite_minmax(leonard)
+         else
+            tnu => dummyvector
+            leonard => dummytensor
          end if
       else
          les_second_order=.false.; les_fourth_order=.false.; wale=.false.; dynamic_les=.false.