[R] .Fortran()

John Bjornar Bremnes J.B.Bremnes at dnmi.no
Wed Oct 10 16:07:52 CEST 2001


Prof Brian Ripley wrote:
> 
> On Tue, 9 Oct 2001, John Bjornar Bremnes wrote:
> 
> > Prof Brian Ripley wrote:
> > >
> > > On Tue, 9 Oct 2001, John Bjornar Bremnes wrote:
> > >
> > > > Prof Brian Ripley wrote:
> > > > >
> > > > > On Mon, 8 Oct 2001, John Bjornar Bremnes wrote:
> > > > >
> > > > > > I am trying to call some Fortran subroutines using .Fortran(). These
> > > > > > subroutines have arguments of type 'character', 'real', 'integer', and
> > > > > > 'integer*2'. Is it impossible to call  subroutines expecting both
> > > > > > 'character' and 'real' as input?
> > > > >
> > > > > No, but it is impossible to call those expecting 'integer*2'
> > > >
> > > > Ok, it seems that I can easily avoid using 'integer*2'
> > > >
> > > > > See ?.Foreign for what is supported, and note the comments there about
> > > > > character types: they work on some compilers and not on others.
> > > >
> > > > Is g77 one of those where it does not work? I use R-1.3.0 on linux (PC).
> > >
> > > A properly working version of g77 should work on some platforms (including
> > > Linux), but not others (including Windows).
> > >
> > > > I have tried the following simple example:
> > > >
> > > >       subroutine mysub(filein, m, x, n)
> > > > c
> > > >       integer        m, n, i
> > > >       character*255  filein
> > > >       real           x(n)
> > > > c
> > > >       write(6,*) 'file: ', filein(1:m)
> > > >       do i=1,n
> > > >          x(i) = sqrt(1.0*i)
> > > >       enddo
> > > >       end
> > > >
> > > > % R CMD SHLIB mysub.f 
> > > > g77   -fPIC  -g -O2 -c mysub.f -o mysub.o
> > > > gcc -shared  -o mysub.so mysub.o -L/usr/local/lib -lg2c -lm
> > > > -L/usr/local/lib -L. -L/usr/lib/gcc-lib/i386-linux/2.95.2 -lm
> > > >
> > > > > dyn.load("mysub.so")
> > > > > .Fortran( "mysub", as.character("myfile"), as.integer(6), as.single(rep(0,10)), as.integer(10), DUP=T )
> > > >  file: myfile
> > > > Segmentation fault (core dumped)
> > >
> > > That works for me (RH6.2, gcc 3.0.1, R 1.3.1).
> > >  file: myfile
> > > [[1]]
> > > [1] "myfile"
> > >
> > > [[2]]
> > > [1] 6
> > >
> > > [[3]]
> > >  [1] 1.000000 1.414214 1.732051 2.000000 2.236068 2.449490 2.645751 2.828427
> > >  [9] 3.000000 3.162278
> > > attr(,"Csingle")
> > > [1] TRUE
> > >
> > > [[4]]
> > > [1] 10
> > >
> > > BTW, no one said Fortran I/O would work, and it sometimes is a problem.
> > >
> > > > % R
> > > > > dyn.load("mysub.so")
> > > > > .Fortran( "mysub", as.character("myfile"), as.integer(6), as.single(rep(0,10)), as.integer(10), DUP=F )
> > > > Error: character variables must be duplicated in .C/.Fortran
> > >
> > > > Any comments?
> > >
> > > Do you have one of the many broken Linux versions of g77?
> > >
> >
> > I have now tried versions 2.95.2 and 3.0 with no success. I will ask my
> > system administrator to install the latest gcc version (3.0.1?). Maybe I
> > should update R to 1.3.1 also. BTW where is the default compiler options
> > set for 'R CMD SHLIB'?
> 
> From the configure script, in etc/Makeconf.
> 
> I doubt if 3.0.1 would help much over 3.0.  It's puzzling though.
> 
> > In
> > "Writing R Extensions" on page 23 it says "It is not possible to pass
> > numeric vectors as float * or REAL if DUP=TRUE". It should be DUP=FALSE,
> > or?
> 
> It should be FALSE, and I've jut fixed it, thanks.

You were right about gcc 3.0.1 - it did not help. 

I then installed R-1.3.1 on a computer running a RH 7.? distribution of
linux with gcc 2.96 and it worked! All my previous tests were on a
Debian distribution. My system administrator thinks the problems may be
related to libc. The debian system use version 2.2.3 of this library and
RH 2.1.9 (if I remember correctly).

I will try out more complex Fortran subroutines (which work fine using
Splus 5.1 on irix 6.5) on the RH system.


thanks again
John Bjornar Bremnes
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._



More information about the R-help mailing list