On Nov 23, 2007, at 11:59 PM, Koun SHIRAI wrote:
> I am looking back with a sigh that there was a long way to reach this
> point. I read the installation manual, and some notices regarding
> machine-dependent procedures. But, it is not easy. In the manual,
> the first command to be issued is as follows,
> shell$ ./configure CC=cc CXX=CC FC=f77 −−prefix=/directory/to/
> install/in
> Most people would do that. Why don't you list concrete configuration
> commands for each machines. Of course, it is impossible to do so for
> all the machines and environments. But, you could do that for
> several cases of major situations. I know that Mac's specific
> configuration was described in sec. 4.4.6. But, the above way is not
> directly written.
We put in all the information we know about a given platform in the
user documentation. But the problem with your suggestion is that
while the documentation is fixed once a release occurs, the platform
is not. In this case, the extra flag is needed because of a change
Apple made to their linker after the last LAM release. Since LAM is
not going to have another release unless a major bug is found, this is
just how things shake out.
> I barely come to know that the LAM implement does not work with
> Fortran 90 in this mailing list, after getting many errors. I am
> disappointed to know this. I am eager to know if it become to work
> with 90 in OpenMPI?
LAM/MPI works with Fortran 90, it just doesn't provide the Fortran 90
module interface. Open MPI does provide a Fortran 90 module.
Brian
--
Brian Barrett
LAM/MPI Developer
Make today a LAM/MPI day!
|
