I intended this check to be a simple sanity check to ensure that we were
getting consistent test results between F77 and F90.  However, it turned
into a visit into what I like to call the Alignment House of Horrors.

The short version is that this test is now:

A) an inaccurate warning that should be removed
B) the result of a long, sordid story

The gist of B) is exactly what the warning message says: the alignment
of a type between F77 and F90 does not match!  This is fairly amazing to
me (indeed, in this case, it is the *same* compiler, but operating in
F77 mode vs. F90 mode), but apparently it's common practice in Fortran
compilers.

We went back and forth about this with various members of the Fortran
committee and various Fortran compiler authors and could never give us a
definite "yes, it will always be this way" kind of answer.  That being
said, we haven't yet run into a place where these alignment problems
have been a problem, even though our tests indicate that the alignments
are wrong.

To be fair, we aren't 100% sure that our F90 test is right (and couldn't
get a definitive answer out of the F90 compiler authors). 

So I'm pretty sure that we left it that this test is fairly meaningless.
I'll do a little more checking and will probably end up removing it (may
take a few days because everyone is traveling home from the OMPI
Developer's Workshop and I need to check with a few of the other
developers).

Thanks for reminding me -- I had intended to close this out a while ago
and forgot about it.




> -----Original Message-----
> From: users-boun...@open-mpi.org 
> [mailto:users-boun...@open-mpi.org] On Behalf Of Michael Kluskens
> Sent: Thursday, April 20, 2006 12:19 PM
> To: Open MPI Users
> Subject: [OMPI users] OMPI-F90-CHECK macro needs to be updated?
> 
> Getting warnings like:
> 
> WARNING: *** Fortran 77 alignment for INTEGER (1) does not match
> WARNING: *** Fortran 90 alignment for INTEGER (4)
> WARNING: *** OMPI-F90-CHECK macro needs to be updated!
> 
> same for LOGICAL, REAL, COMPLEX, INTEGER*2, INTEGER*4, INTEGER*8, etc.
> 
> I believe these are new within the last week.
> 
> The configure script is generating these warnings when building  
> openmpi-1.1a3r9663 using g95 from April 19 on OS X 10.4.6.   
> Configured with "./configure F77=g95 FC=g95 LDFLAGS=-lSystemStubs"
> 
> 

Reply via email to