[R-sig-Fedora] problems compiling for RHEL 5.3 x86_64

[egc]

Hei Roger - thanks - comments below...

Roger Bivand wrote:
> On Fri, 26 Jun 2009, Evan Cooch wrote:
>
>>
>> So, several quick questions:
>>
>> 1. anyone managed to get x86_64 R compiled under either RHEL, Fedora, 
>> or CentOS? I don't mean 'installed from an RPM', but...compiled from 
>> source.
>
> Yes, RHEL5, x86_64, no problem.

Since CentOS 5.3 is 99.95% RHEL 5.3, I guess it amounts to the same 
thing - or it should. 

>
>>
>> 2. anyone who answers 'yes' to (1) - did you manage to get it to play 
>> nice with blas and lapack?
>
> Yes, with GoToBLAS, using the instructions in:
>
> http://cran.r-project.org/doc/manuals/R-admin.html#Linear-algebra
>
> especially:
>
> "Another option to change the BLAS in use is to symlink a dynamic BLAS 
> library (such as ACML or Goto's) to R_HOME/lib/libRblas.so. For 
> example, just
>
>      mv R_HOME/lib/libRblas.so R_HOME/lib/libRblas.so.keep
>      ln -s /opt/acml4.0.1/gfortran64_mp/lib/libacml_mp.so \
>      R_HOME/lib/libRblas.so
>
> will change the BLAS in use to multithreaded ACML."
>
>
I tried this, but got hung up having to set some environment variables, 
and never seemed to get it to work. I didn't try GoToBLAS, but the ACML 
blas.

Where I'm still somewhat puzzled is the lapack end of things. blas is 
the basic lib, but lapack makes it zippy fast. Some blas has lapack in 
it (like ACML). But, R seems to not to like compiling against it. As 
long as I use

./configure --with-tcltk

make, make install, then change the link to libRblas.so (as noted 
above), no problem - except that R uses the liblapack.so, which it 
didn't create from ACML (since I didn't use the 
--with-blas=/pointer/to/ACML -lacml  directive during config. If I do 
this, however, using ./configure --with-tcltk --with-lapack 
--with-blas="-L/pointer/to/ACML/lib -lacml", the make check fails 
miserably on anything requiring high-level algebra).