[MOAB-dev] Disable fortran in moab-5.0.0

[Vijay S. Mahadevan]

Paul, there is another reason why I added the
--with-fc-mangling-suffix option. This is for the corner case when you
want to use external BLAS/LAPACK, without Eigen and without Fortran
support. This is the only way for us to decipher what the mangling
should be. Of course, it only covers the case of no underscore, one or
double underscore suffixes. Not any changes to case etc.

Anyway, hope that gives another reason why we need the changes related
to this. On a separate note, none of our builds on develop have broken
yet and so I'll submit a PR for this tomorrow. If there are other
comments or feature requests, let me know.

Vijay

On Tue, Nov 14, 2017 at 11:05 PM, Vijay S. Mahadevan <vijay.m at gmail.com> wrote:
> Hi Paul,
>
> There are a couple of protos (CN/ that get included in some source
> files. I haven't fully tested this branch on all our builds yet but if
> all of them do pass, I believe we may not need mangling. We are of
> course talking about the case of not having a fortran compiler,
> whether you want BLAS or not.
>
> Vijay
>
> On Tue, Nov 14, 2017 at 10:34 PM, Paul Wilson <paul.wilson at wisc.edu> wrote:
>> Hi Vijay,
>>
>> Maybe I'm missing something, but if building without BLAS/LAPACK, what is
>> the mangling still needed for?
>>
>> Paul
>>
>>
>>
>> On 11/14/2017 07:58 PM, Vijay S. Mahadevan wrote:
>>>
>>> Hi Lorenzo,
>>>
>>> Can you give the branch "vijaysm/osx-blas-lapack" a try on your
>>> machines ? I've added several additional configure options to control
>>> the mangling (when fortran is disabled), and an option to disable
>>> BLAS/LAPACK (only if Eigen is present). A combination of:
>>> --disable-blaslapack --with-eigen3 --with-fc-mangling-suffix="_"
>>> should give you the right result. Let me know if this gives you the
>>> result of not requiring a Fortran compiler.
>>>
>>> I will merge this branch to develop and run our buildbot tests on
>>> various configuration combinations. Paul, I'll also make some
>>> modifications to CMake if this approach is acceptable. Let me know.
>>>
>>> Iulian, any comments ?
>>>
>>> Vijay
>>>
>>> On Fri, Nov 10, 2017 at 5:14 PM, Vijay S. Mahadevan <vijay.m at gmail.com>
>>> wrote:
>>>>
>>>> Paul, as Lorenzo mentioned, we need to modify configuration to enable
>>>> this cross option. I can look into that as well.
>>>>
>>>> The more combinations/options we add, the more complicated the
>>>> configuration becomes. While things do become seemingly flexible, I
>>>> just want to make sure it doesn't become a maintenance nightmare later
>>>> on. Hence the decision originally to just make BLAS/LAPACK mandatory
>>>> and not allow Eigen (with/without external BLAS). The implication on
>>>> FC here was a side effect unfortunately of that decision.
>>>>
>>>> Vijay
>>>>
>>>> On Fri, Nov 10, 2017 at 3:29 PM, Lorenzo Botti <bottilorenzo at gmail.com>
>>>> wrote:
>>>>>
>>>>> Dear Paul,
>>>>> I don’t think it’s actually possible. The build system should be
>>>>> modified to
>>>>> account for such a configure option.
>>>>> Anyway this is also what I need, a fix to run MOAB on OSX without
>>>>> fortran.
>>>>> On my laptop I’m not interested on performance but I do most of the
>>>>> development.
>>>>> Installing a fortran compiler on OSX is a pain, as far as I know this
>>>>> forces
>>>>> me to use gcc instead of clang.
>>>>> Accordingly I would need to use a package manager like macports or
>>>>> homebrew,
>>>>> or to compile gcc myself.
>>>>> A fallback on Eigen would be really welcome.
>>>>>
>>>>> Bests
>>>>> Lorenzo
>>>>>
>>>>> On 10 Nov 2017, at 21:19, Paul Wilson <paul.wilson at wisc.edu> wrote:
>>>>>
>>>>> I'm not sure if I follow how to actually do what you are suggesting.  If
>>>>> I
>>>>> wanted to rely on Eigen instead of both BLAS/LAPACK, is there a set of
>>>>> options I can pass to either autoconf or CMAKE to do that?
>>>>>
>>>>>
>>
>> --
>> -- ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ --
>>
>> Paul P.H. Wilson
>> Grainger Professor of Nuclear Engineering
>> 608-263-0807
>> paul.wilson at wisc.edu
>> 443 Engineering Research Bldg
>> 1500 Engineering Dr, Madison, WI 53706
>> calendar: http://go.wisc.edu/pphw-cal
>>
>> Computational Nuclear Engineering Research Group
>> cnerg.engr.wisc.edu
>>