C/F blas/lapack (tying to build in cygwin)

[Lisandro Dalcin]

On Sat, Mar 21, 2009 at 8:05 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
>
>   The cblaslapack binding does not, in general, match any particular Fortran
> compiler, thus calling cblaslapack from some Fortran compiler might not
> work. To prevent this situation we have the check that you hit below.
>

OK, I see... it make sense...

>   It would be good to figure out why gfortran could not handle the
> f-blas-lapack and fix it, please send us the configure.log file if you still
> have it.
>

I suspect the cygwin gfrotran from GCC4 is perhaps a bit datated, but
I'll need to confirm this (I'm using all this windows crap at home).
I'll try again and send the configure log...

>
> One could argue that we shouldn't worry about this and let the user screw
> themselves but generally our model is to make it difficult for PETSc users
> to screw themselves.
>

OK, now I agree this is a good thing ...

>
>
>
>> The theory is when fortran compiler exists - f-blas-lapack should
>> always work.
>>
>> And when a fortran compiler exists - its name-mangling might not match
>> the cblas namemangling... [so clblaslapack with fc is disabled]
>>
>> Satish
>>
>> On Sat, 21 Mar 2009, Lisandro Dalcin wrote:
>>
>>> I was trying to build petsc-dev on win32+cygwin using GCC 4. For some
>>> reason I did not investigate, gfortran failed to build f-blas-lapack,
>>> then re-ran configure like below. Why you do not let me use a
>>> c-blas-lapack despite the fact that I have a Fortran compiler?
>>>
>>> $ python config/configure.py --with-cc=gcc --with-fc=gfortran
>>> --download-c-blas-lapack=1 --with-mpi=0
>>>
>>> =================================================================================
>>>            Configuring PETSc to compile on your system
>>>
>>> =================================================================================
>>> TESTING: configureLibrary from
>>>
>>> config.packages.BlasLapack(config/BuildSystem/config/packages/BlasLapack.py:429)
>>>
>>>
>>> *********************************************************************************
>>>        UNABLE to CONFIGURE with GIVEN OPTIONS    (see configure.log
>>> for details):
>>>
>>> ---------------------------------------------------------------------------------------
>>> Should request f-blas-lapack, not --download-c-blas-lapack=yes since
>>> you have a fortran compiler?
>>>
>>> *********************************************************************************
>>>
>>>
>>>
>>>
>>>
>>>
>>
>
>



-- 
Lisandro Dalcín
---------------
Centro Internacional de Métodos Computacionales en Ingeniería (CIMEC)
Instituto de Desarrollo Tecnológico para la Industria Química (INTEC)
Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET)
PTLC - Güemes 3450, (3000) Santa Fe, Argentina
Tel/Fax: +54-(0)342-451.1594