We would like to compile a minimal R library that could be linked into an application that will be run on a BlueGene/L system with 8,192 processors. This is a system that requires no shared libraries, no graphical interface, must be single threaded, and will be cross- compiled. I would statically link the code for the packages we require. From looking through the code, it seems like it should be possible to build such a version. However, it would be invaluable to have the advice and aid of the people who truly know this software. Thus far, I have configured the build to use the appropriate compilers. I had to tell configure that it was crosscompiling by editing the config.site file (it didn't recognize that it was). However, I am currently hitting a roadblock with configure failing with the report: WARNING: blrts_xlf and blrts_xlc disagree on int and double configure: error: Maybe change CFLAGS or FFLAGS The last lines in config.log are: #ifdef __cplusplus extern "C" void exit (int) throw (); configure: exit 1 Thank you very much for any assistance, Best Sean
Quoting the last few lines of config.log is not useful - you need to post the whole of it (or let it be downloaded from somewhere), since the error is a good deal further up. The error and fix is as it says - it seems that your fortran compiler (blrts_xlf) disagree with your C compiler (blrts_xlf) about how big is int and double (On most usual boxes these days int is 32-bit and double is 64-bit), and you will need to read your compiler's manual about what size are used for int and double accordingly; this is highly compiler-specific so you'll have to read it up yourself. Sean Hill wrote:> We would like to compile a minimal R library that could be linked > into an application that will be run on a BlueGene/L system with > 8,192 processors. This is a system that requires no shared libraries, > no graphical interface, must be single threaded, and will be cross- > compiled. I would statically link the code for the packages we require. > > From looking through the code, it seems like it should be possible > to build such a version. However, it would be invaluable to have the > advice and aid of the people who truly know this software. > > Thus far, I have configured the build to use the appropriate > compilers. I had to tell configure that it was crosscompiling by > editing the config.site file (it didn't recognize that it was). > However, I am currently hitting a roadblock with configure failing > with the report: > WARNING: blrts_xlf and blrts_xlc disagree on int and double > configure: error: Maybe change CFLAGS or FFLAGS > > The last lines in config.log are: > #ifdef __cplusplus > extern "C" void exit (int) throw (); > > configure: exit 1 > > Thank you very much for any assistance, > Best > Sean > > ______________________________________________ > R-devel at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel
On Mon, 19 Jun 2006, Sean Hill wrote:> > We would like to compile a minimal R library that could be linked > into an application that will be run on a BlueGene/L system with > 8,192 processors. This is a system that requires no shared libraries, > no graphical interface, must be single threaded, and will be cross- > compiled. I would statically link the code for the packages we require. > > From looking through the code, it seems like it should be possible > to build such a version. However, it would be invaluable to have the > advice and aid of the people who truly know this software. > > Thus far, I have configured the build to use the appropriate > compilers. I had to tell configure that it was crosscompiling by > editing the config.site file (it didn't recognize that it was).There is no support for cross-compiling R (except for Windows on Linux/...). This may explain the test failure below, as it runs an executable.> However, I am currently hitting a roadblock with configure failing > with the report: > WARNING: blrts_xlf and blrts_xlc disagree on int and double > configure: error: Maybe change CFLAGS or FFLAGS > > The last lines in config.log are: > #ifdef __cplusplus > extern "C" void exit (int) throw (); > > configure: exit 1 > > Thank you very much for any assistance, > Best > Sean > > ______________________________________________ > 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