[MITgcm-support] new compile fails

[Michael Schaferkotter (Contractor)]

greetings:
after having done numerous successful compile, setup, and run MITgcm experiments under MPI 
(even as recently as late jan), i know get a compile error on a current build and also
the same error upon rebuilding previously working experiments.

here/s the compile error:

        cat mdswritefield_new.F | $MITgcmHOME/tools/set64bitConst.sh |  /lib/cpp -P -DALLOW_USE_MPI -DALWAYS_USE_MPI -DTARGET_AIX -DWORDLENGTH=4 -DALLOW_USE_MPI -DALWAYS_USE_MPI -DHAVE_SYSTEM -DHAVE_SETRLSTK -DHAVE_SIGREG -DHAVE_STAT 
-I/usr/lpp/ppe.poe/include/thread64 -I/usr/local/include > mdswritefield_new.f
        mpxlf95_r -q64 -qfixed=132 -O3 -Q -qarch=pwr4 -qtune=pwr4 -qcache=auto -qmaxmem=-1 -c mdswritefield_new.f
"mdswritefield_new.f", line 2104.25: 1514-006 (S) Variable mpi_myxgloballo is already an element of a COMMON block.  Variable is ignored.
"mdswritefield_new.f", line 2104.42: 1514-006 (S) Variable mpi_myygloballo is already an element of a COMMON block.  Variable is ignored.
"mdswritefield_new.f", line 2105.15: 1514-071 (W) Identifier mpi_myxgloballo was previously defined with same type.
"mdswritefield_new.f", line 2106.15: 1514-071 (W) Identifier mpi_myygloballo was previously defined with same type.
** mdswritefield_new   === End of Compilation 1 ===
1501-511  Compilation failed for file mdswritefield_new.f.
make: 1254-004 The error code from the last command is 1.


here/s the top of the Makefile

#    Tue Mar 28 18:28:43 GMT 2006# Multithreaded + multi-processing makefile for:
#    AIX r4n3 2 5 0025617C4C00
# This makefile was generated automatically on
#    Tue Mar 28 18:28:43 GMT 2006
# by the command:
#    $MITgcmHOME/tools/genmake2  "-mods=../code" "-disable=gmredi,kpp,zonal_filt,shap_filt" "-mpi" "-optfile=$MITgcmHOME/
tools/build_options/sp4+mpi"
# executed by:
#    mich at r4n3.navo.hpc.mil:$MITgcmHOME/MITgcm-expt/ALEX/expt_01.2/build
#
# OPTFILE="$MITgcmHOME/tools/build_options/sp4+mpi"

not being sure if the distribution was up-to-date, i updated from the cvs repository, started over, and
received exactly the same failure message.

looking in the soure code in mdswritefield_new.f, i do find mpi_myxgloballo in two COMMON blocks.

any ideas about all of this?

by the way, by adding '-qhalt=q' to the flags gets past all of this, but that flag  is 'do not halt, even for unrecoverable errors',
the '-qhalt=u' still fails (this is quit on unrecoverable error messages).

m. r. schaferkotter