fOptions no longer builds: Fortran error

[Martin Maechler]

Thank you, Dirk,

>>>>> Dirk Eddelbuettel <edd at debian.org>
>>>>>     on Tue, 7 Jun 2011 08:55:27 -0500 writes:

    > Christophe,

    > Would you be able to help in this Fortran-related build failure
    > for fOptions?  The compiler now complains that LowDiscrepancy
    > has conflicting types, and with my limited Fortran skills I see
    > no fix -- the declarations seem fine to me.

    > The URL below has more detail, and I'd be happy to fill in more
    > details.

    > Thanks,  Dirk

> On 28 May 2011 at 16:58, Dirk Eddelbuettel wrote:
> | 
> | Hi all,
> | 
> | During a routine rebuild of the Debian archive, it was noticed that
> | fOptions no longer builds (http://bugs.debian.org/628314) as the Fortran
> | compiler (now gfortan-4.6) seems pickier:
> | 
> | > make[1]: Entering directory `/build/user-foptions_2110.78-1-amd64-xexJr_/foptions-2110.78/src'
> | > gfortran   -fpic  -O3 -pipe  -g -c 085A-LowDiscrepancy.f -o 085A-LowDiscrepancy.o
> | > 085A-LowDiscrepancy.f:181.32:
> | > 
> | >                QN(I, J) = HQNORM(QUASI(J))                              
> | >                                 1
> | > Error: Return type mismatch of function 'hqnorm' at (1) (REAL(4)/REAL(8))
> | > make[1]: *** [085A-LowDiscrepancy.o] Error 1
> | 
> | I do not write Fortran so would not know how to fix this. Any RMetrics export
> | who can fix this?

would 'export' be 'expert' ?

As you say somewhere, we may rather need an extreme Fortran
expert, as indeed the Fortran source's declaration seem ok, to
me as well.

Can we be 99.99% sure that this is *not* a compiler bug/error ?

My approach in such cases would be to translate to C 
and continue working with that...

Martin



> | 
> | Dirk
> | 
..