[R] 64-bit Windows 2003 build of R

[Peter Dalgaard]

"John Sweval" <JSweval at illumigen.com> writes:

> I am running into serious memory constraint issues with the 32 bit build
> of R.  We have Windows 2003 on an Itanium 64 box with the Intel 64-bit C
> compiler (8.0.041) and want to create a 64-bit version of R.
> Unfortunately, I am no longer "current" with C builds, i.e. - I haven't
> done one in years and have never used the Intel 64-bit compiler.  I
> could use any suggestions, make file, assistance, etc. from anyone who
> may have already compiled a 64-bit Windows version of R.

Well, as you probably realized while typing that, you'll likely be
short of luck unless you find someone with similar hardware, OS, and
compilers PLUS the stamina to go through the "flagmire" of configuring
for a different set of compilers/linkers than those normally used to
build R on Windows. And 64-bit Windows installs appear to be quite
rare. 

It's not easy. Building R with other compilers than the recommended
GCC/MinGW tools is hard (as in "not recommended"!) even on 32 bit
Windows.

Then there's the option of ignoring the Intel compiler and using GNU
tools like the ones for Win32, but I suspect that you'll still run
into porting issues. It shouldn't be too hard to cross-build a 64 bit
GCC compiler and binutils toolchain, but there's still an issue about
what to do with the win32api stuff.

The most efficient way to get a 64 bit system running here and now is
to switch the operating system entirely, since R is known to build
pretty much out of the box on Linux for IA64 and x86-64 systems.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907