[Rd] Problems building R on Solaris (gcc 4.6, 64 bit)

Chander Ganesan chander at otg-nc.com
Tue Jun 5 16:48:55 CEST 2012


On 6/5/12 9:38 AM, Prof Brian Ripley wrote:
> On 04/06/2012 19:27, Chander Ganesan wrote:
>> Hello,
>>
>> I'm attempting to build R on Solaris 64-bit using gcc4, and am running
>> into the following error:
>>
>> ld.so.1: R: fatal: relocation error: R_AMD64_PC32: file
>> /home/chander.ganesan/src/r/R-2.15.0/lib/libR.so: symbol main: value
>> 0x28001597b8c does not fit
>>
>>
>> I've noticed - from some lists, and a bug here
>> (http://trac.sagemath.org/sage_trac/ticket/9040) that this isn't an
>> unexpected issue - but I'm wondering if anyone has a fix for it? If not,
>> would someone be willing to develop a permanent fix for it (which I
>> believe I could fund, within reason.)
>>
>> Thank You
>
> R's own documentation has had workarounds for some time: please study 
> the R-admin manual.  I built R-patched using 4.7.0 this morning on 
> both Solaris 64-bit) platforms (x64 and Sparc).
I did look there - and followed the instructions therein (followed by 
numerous variants of it based on google searches, etc.) I don't recall 
seeing any mention of R-patched in the documentation, but I suppose I 
could have missed that somehow.  IIRC that document indicates that R 
doesn't build 64-bit using gcc, but I also couldn't get it to work with 
Sun Studio 12.3 compilers.
>
> You do not say what version of R you are trying: I vaguely recall an 
> improvement here since 2.15.0 so suggest you try R-patched (and read 
> its manual).  It also looks as if you are building R as a shared 
> library, which is not the default and it might be better to try the 
> defaults first.
I was using 2.15 - but with R_patched I'm still having some issue.  
Could you possibly send me the arguments you are using with configure to 
build on solaris x64 ?

Thank You

-- 
Chander Ganesan
Open Technology Group, Inc.
11010 Lake Grove Blvd Ste. 100-307
Morrisville, NC  27560
919-463-0999/877-258-8987
http://www.otg-nc.com



More information about the R-devel mailing list