Gentle Reader,

Certain F77 compilers -- Compaq DEC, Sun, and perhaps others -- will
not allow the MAX or MIN intrinsics to be used in the declaration
section of a Fortran routine that uses IMPLICIT NONE because they are
not typed (even though they are intrinsics).  One example of this is
in the MM5v3.1 routine domain/boundary/bdyin.F :

  REAL SCR(MAX(MIX,MIX_G)*MAX(MIX,MJX_G)*MAX(NSPGD,NSPGX))      BDYIN.40

Another is in domain/io/outtap.F :

  REAL  HDOTH(MAX(MIX,MIX_G)*MAX(MJX,MJX_G)*(MKX+1))            OUTTAP.79

There are a couple of works-around for this.  You can remove IMPLICIT
NONE on the routine (although that is not really desireable,
considering the benefits of IMPLICIT NONE) or you can simply put a line
to type the MAX intrinsic prior to the offending lines, above:

  INTEGER MAX

Another work around would be to modify your configure.user file to use
f90, instead.

The f77 compiler appears unable to recognize that MAX is an intrinsic
function and ignore the fact that it's not typed (as it does in the
executable part of the code). It's not clear, however, that this is a
compiler bug because IMPLICIT NONE itself is not standard F77.

-Rotang