[Rd] The standalone Rmath library and VC++ 2003

Prof Brian Ripley ripley at stats.ox.ac.uk
Wed Jan 30 19:35:47 CET 2008


On Wed, 30 Jan 2008, Oran Johnson wrote:

> Yes, I used Rtools to compile the standalone Rmath library. However,
> C:\Rtools\MinGW\lib\libm.a does not contain log1p;
> C:\Rtools\MinGW\lib\libmingwex.a does. Once I replaced libm.a with
> libmingwex.a, the application linked successfully and the following test
> program was successful:

MinGW's loader will pick up the correct set of default libraries, which 
does include -lmingwex and much else, so if you build a DLL, these entry 
points are compiled in already.

Otherwise, it is down to you to figure out how to link, and it may change 
from time to time. Just the same experience as when you mix-and-match 
compilers on a Unix-alike (try mixing gcc3 and gcc4, or gcc and Sun 
Studio).


>
> #include <iostream>
> #define MATHLIB_STANDALONE
> #include <Rmath.h>
>
> int main()
> {
> 	double d = dbeta(0.5, 2.0, 3.0, 1);
> 	double q = qbeta(0.5, 2.0, 3.0, 1, 0);
> 	double p = pbeta(0.5, 2.0, 3.0, 1, 0);
> 	double r = rbeta(0.5, 2.0);
>
> 	std::cout << "dbeta = " << d << "\n";
> 	std::cout << "qbeta = " << q << "\n";
> 	std::cout << "pbeta = " << p << "\n";
> 	std::cout << "rbeta = " << r << "\n";
> }
>
>
> dbeta = 0.405465
> qbeta = 0.385728
> pbeta = 0.6875
> rbeta = 0.422286
> Press any key to continue
>
> Thanks
> Oran
>
> -----Original Message-----
> From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
> Sent: Wednesday, January 30, 2008 1:06 PM
> To: Oran Johnson
> Cc: r-devel at r-project.org
> Subject: Re: [Rd] The standalone Rmath library and VC++ 2003
>
> On Tue, 29 Jan 2008, Oran Johnson wrote:
>
>> Linking my VC++ application with the standalone Rmath library yields
> the
>> following;
>
> Looks like you compiled the standalone Rmath library with MinGW.  You
> can't expect to mix-and-match compilers and link together a static
> library
> built under another compiler.
>
> It is possible that this will work if you try linking to Rmath.dll:
> WIndows' DLLs are fairly portable (although parameter passing
> conventions
> do differ, and we know 'double complex' is not portable).  I assume you
> know how to make a VC++ import library from a DLL, which you would need
> to
> do.
>
> If you compiled standalone Rmath under VC++, you would not have got a .a
>
> but a lib, at least by default.  You are welcome to try that: it has
> been
> done in the past but is not something we support (nor even would want to
>
> answer questions about).
>
>
>> ------ Build started: Project: Complex plugin, Configuration: Debug
>> Win32 ------
>>
>>
>>
>> Linking...
>>
>>   Creating library .\../Debug/complex_plugin.lib and object
>> .\../Debug/complex_plugin.exp
>>
>> libRmath.a(mlutils.o) : warning LNK4217: locally defined symbol __iob
>> imported in function _REprintf
>>
>> libRmath.a(dbeta.o) : error LNK2019: unresolved external symbol _log1p
>> referenced in function _dbeta
>>
>> libRmath.a(qbeta.o) : error LNK2019: unresolved external symbol _log1p
>> referenced in function _qbeta
>>
>> libRmath.a(lbeta.o) : error LNK2019: unresolved external symbol _log1p
>> referenced in function _lbeta
>>
>> libRmath.a(toms708.o) : error LNK2001: unresolved external symbol
> _log1p
>>
>> libRmath.a(qbeta.o) : error LNK2019: unresolved external symbol _expm1
>> referenced in function _qbeta
>>
>> libRmath.a(mlutils.o) : error LNK2019: unresolved external symbol
>> ___fpclassify referenced in function _R_pow
>>
>> ../Debug/complex_plugin.dll : fatal error LNK1120: 3 unresolved
>> externals
>>
>>
>>
>> Build log was saved at "file://c:\starting point\Debug\BuildLog.htm"
>>
>> Complex plugin - 7 error(s), 1 warning(s)
>>
>>
>>
>>
>>
>> ---------------------- Done ----------------------
>>
>>
>>
>>    Build: 0 succeeded, 1 failed, 0 skipped
>>
>>
>>
>> At first glance, it looks like C:\Rtools\MinGW\lib\libm.a is missing.
>> After adding it to my project properties, the results are the same. It
>> looks like VC++ is OK with GNU libraries because it appears to accept
>> libRmath which was built on the same machine using Rtools. Also it
> does
>> not appear to be a configuration issue, because it's only found
>> 'C:\Rtools\MinGW\lib' and VC+ is complaining about not finding it.
>>
>>
>>
>> Any ideas?
>>
>>
>>
>> Oran
>>
>>
>>
>>
>> 	[[alternative HTML version deleted]]
>>
>> ______________________________________________
>> R-devel at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>
>
> --
> Brian D. Ripley,                  ripley at stats.ox.ac.uk
> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
> University of Oxford,             Tel:  +44 1865 272861 (self)
> 1 South Parks Road,                     +44 1865 272866 (PA)
> Oxford OX1 3TG, UK                Fax:  +44 1865 272595
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595



More information about the R-devel mailing list