[Rd] problem running dll in R

Duncan Murdoch murdoch at stats.uwo.ca
Mon Aug 25 20:13:37 CEST 2008


On 8/25/2008 11:45 AM, rcoder wrote:
> Hi Duncan,
> 
> Thanks for your reply. Would you know of a code + compiler combination that
> works in R, so I can get a feel for the process?

Sure, the MinGW compilers in the Rtools should be fine.  You get them 
from www.murdoch-sutherland.com/Rtools.  With those, building a DLL is 
simply

R CMD SHLIB source.f

Duncan Murdoch

> 
> Thanks,
> 
> rcoder
> 
> 
> 
> Duncan Murdoch-2 wrote:
>> 
>> On 23/08/2008 8:44 PM, rcoder wrote:
>>> Hi everyone,
>>> 
>>> I've created a dll for a fortran subroutine, and when I call the
>>> subroutine
>>> in R (via a wrapped function) a standard program shutdown windows error
>>> prompt immediately pops up: the program has encountered a problem and
>>> needs
>>> to be shut down... send message / don't send message to MS... etc.
>>> 
>>> When I click "Don't send", the R consol does not then shut down, and
>>> appears
>>> to continue to execute, and task manager does not indicate "not
>>> responding".
>>> The problem is the execution takes too long, and I think the program is
>>> just
>>> hanging rather than executing. I have tested with different fortran code,
>>> and the same thing happens, so I don't think it is anything to do with
>>> that
>>> side of things.
>>> 
>>> I created fortran code in a text editor, compiled using Plato3 IDE, then
>>> converted the file into a dll using gFortran. I place the dll in the R
>>> directory and it load in the console. The problem seems to occur when
>>> defining the data types in R, for the variables in the fortran routine.
>>> 
>>> I was just wondering if anyone has experienced anything similar when
>>> trying
>>> to call fortran subroutines in R?
>> 
>> I haven't seen that, and haven't used that particular compiler, but it 
>> has the symptoms of a mismatch in calling conventions.  Does your 
>> compiler pass arguments in registers?  R uses what Windows calls the 
>> cdecl calling convention in external calls.
>> 
>> Some old advice on using non-standard compilers is available here:
>> 
>> http://www.stats.uwo.ca/faculty/murdoch/software/compilingDLLs
>> 
>> I haven't updated it in a few years, but I think it is still current.
>> 
>> Duncan Murdoch
>> 
>> ______________________________________________
>> R-devel at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
>> 
>> 
>



More information about the R-devel mailing list