[R] R CMD SHLIB changes function name when compiling

[Steve Pederson]

It seems that was the problem. My apologies as I was unaware of the 
difference between a .c file & a .C file.

Thanks again for your help,

Steve



On 9/10/2010 1:37 AM, Prof Brian Ripley wrote:
> Is the file .c or .C?  You said .c, but then you said a .C file worked
> (which it should not have), and I wonder if you have this backwards.
>
> A .c file is C, a .C file is C++ (in R 2.11.1 but not R 2.12.0).
>
> extern "C" would be required if this were a C++ source, but we were told
> otherwise.
>
> On Sat, 9 Oct 2010, Steve Pederson wrote:
>
>> Hi Prof Ripley,
>>
>> Thanks for the response & the use of the correct terminology. I've
>> checked the environment variables & there is none set called CC.
>>
>> The .dll was definitely compiled using R CMD SHLIB. I'm not quite sure
>> what output you're referring to but at the command prompt when I enter:
>>
>> ...\bin> R CMD SHLIB summariseMCMC.C
>>
>> I get the output:
>>
>> g++ -I"C:/R/R-2.11.1/include" -02 -Wall summariseMCMC.c summariseMCMC.o
>> g++ -shared -s -static-libgcc -o summariseMCMC.dll tmp.def
>> summariseMCMC.0 -LC:/R/R-2.11.1/bin -1R
>
> Exactly that -- g++ is a C++ compiler.
>
>> Please let me know if I can try anything different provide any more
>> useful output.
>>
>> Another source file I have (runUniformMCMC.C) compiles & runs
>> perfectly using the same approach, so I'm rather perplexed as to why
>> this one gets it's name mangled. It originally had actual code in it
>> that did useful things, but I stripped it back while I was trying to
>> figure out what is happening.
>>
>> Thanks again,
>>
>> Steve
>>
>>
>> On 8/10/2010 8:06 PM, Prof Brian Ripley wrote:
>>> On Fri, 8 Oct 2010, Steve Pederson wrote:
>>>
>>>> Hi,
>>>>
>>>> I'm trying to write a function in C for implementation with .Call.
>>>> When compiling using R CMD SHLIB characters seem to be added to the
>>>> function name.
>>>>
>>>> Here's the complete C code from the file summariseMCMC.c:
>>>>
>>>> #include<R.h>
>>>> #include<Rdefines.h>
>>>> #include<Rinternals.h>
>>>>
>>>> void summariseMCMC(SEXP data) {
>>>>
>>>> PROTECT(data=AS_NUMERIC(data));
>>>> UNPROTECT(1);
>>>>
>>>> }
>>>>
>>>> Then after compiling (R CMD SHLIB summariseMCMC.c) & loading the .dll
>>>>
>>>> dyn.load("C:/R/R-2.11.1/bin/summariseMCMC.dll")
>>>> is.loaded("_Z13summariseMCMCP7SEXPREC")
>>>> [1] TRUE
>>>> is.loaded("summariseMCMC")
>>>> [1] FALSE
>>>>
>>>> Just wondering if anyone had any pointers for getting rid of this, or
>>>> have I missed something outrageously obvious?
>>>
>>> You have. This was not done by'R CMD SHLIB', but by a C++ compiler -- it
>>> is called 'name mangling'. Unfortunately you didn't show us the output
>>> from that command, when the cause would probably have been 'outrageously
>>> obvious'.
>>>
>>> The fix is to make sure you use a C compiler to compile C code, and
>>> we've almost no idea why that is not being done on your system.
>>> But as a guess, check that the environment variable CC is not set.
>>>
>>>>
>>>> Thanks,
>>>>
>>>> Steve
>>>>
>>>>
>>>> sessionInfo()
>>>> R version 2.11.1 (2010-05-31)
>>>> i386-pc-mingw32
>>>>
>>>> locale:
>>>> [1] LC_COLLATE=English_Australia.1252 LC_CTYPE=English_Australia.1252
>>>> [3] LC_MONETARY=English_Australia.1252 LC_NUMERIC=C
>>>> [5] LC_TIME=English_Australia.1252
>>>>
>>>> attached base packages:
>>>> [1] stats graphics grDevices utils datasets methods base
>>>>
>>>> other attached packages:
>>>> [1] limma_3.4.4 Biobase_2.8.0 aroma.affymetrix_1.7.0
>>>> [4] aroma.apd_0.1.7 affxparser_1.20.0 R.huge_0.2.0
>>>> [7] aroma.core_1.7.0 aroma.light_1.16.0 matrixStats_0.2.1
>>>> [10] R.rsp_0.3.6 R.cache_0.3.0 R.filesets_0.8.3
>>>> [13] digest_0.4.2 R.utils_1.4.4 R.oo_1.7.3
>>>> [16] R.methodsS3_1.2.0
>>>>
>>>> loaded via a namespace (and not attached):
>>>> [1] tools_2.11.1
>>>>
>>>> ______________________________________________
>>>> R-help at r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>> PLEASE do read the posting guide
>>>> http://www.R-project.org/posting-guide.html
>>>> and provide commented, minimal, self-contained, reproducible code.
>>>>
>>>
>>
>