Skip to content

Neophyte question re using dyn.load and accessing c functions

3 messages · Barnet Wagman, Robert Gentleman, Brian Ripley

Barnet Wagman
# 
I'm have what's probably a trivial &| silly problem trying to access a
c-function in a shared
library ( R 0.64, under redhat linux 5.2).

    R COMPILE rctest.c
    R SHLIB -o libRC.so rctest.o

produce a reasonable looking shared library ('nm libRC.so' shows that
the function 'rctest' is present).

In R, the command

    dyn.load("/fullpathname/libRC.so")

returns without an error.  However, I'm unable to access the function.
E.g.

    is.loaded(symbol.C("rctest"))

yields FALSE.

    .C("rctest",x)

yields "Error in .C("rctest", x) : C/Fortran function not in load table"

There must be something I'm leaving out, but after searching the
documentation and
help archive, I'm stuck.  Does the .so have to be in an R package (if
so, where)?

Thanks,

-------------------
Barnet Wagman

wagman at enteract.com
773-645-8369

1361 N. Hoyne
Chicago, IL 60622
--------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://stat.ethz.ch/pipermail/r-help/attachments/19990425/efb04208/attachment.html
Robert Gentleman
#
On Sun, 25 Apr 1999, Barnet Wagman wrote:

            

      
      
      
    

        
Well when this happens to me it's usually because R configured incorrectly
and decided to either prepend (or not) an underscore when it wasn't (or
was) needed. 
Stepping through the code using a symbolic debugger will let you know
what's going on (or you could try recompiling with
HAVE_NO_SYMBOL_UNDERSCORE, set to the correct value for your machine).

robert

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Brian Ripley
    

    

      
      #
    

  


  

      
On Sun, 25 Apr 1999, Robert Gentleman wrote:

        

            

      
      
      
    

        
Just to point out that tests almost nothing (at present). You can have
a library that loads with unresolvable symbols that will throw you out of
R when you try to use the symbol.  About the only think it tests is the
presence of the library file.

            

      
      
      
    

        
You can test if the configuration is right by

library(mva)
example(cmdscale)

If that fails it is the system configuration: if it works it is your
example, about which we will need to know more to help....

(This is such a common system for R it is hard to imagine that
configure did not work, but there is always a first time.)