[Rd] Check errors using R2.6.2

[Prof Brian Ripley]

On Mon, 10 Mar 2008, Uwe Ligges wrote:

>
>
> Rossi, Peter E. wrote:
>> I can successfully "check" a package with source under 2.5.1, including
>> compiling source files and running examples with no errors or warnings.
>>
>> when I try with R2.6.2, I get make errors:
>>
>> making bayesmc.d from bayesmc.c
>> make[3]:gcc-sjlj: Command not found
>>
>> etc.
>>
>> my gcc is version 3.4.2
>>
>> I'm using Windows XP.
>>
>> Any thoughts?
>
> Yes: Upgrade gcc as the R Administration and Installation manual
> suggets. R >= 2.6.0 is built under gcc-4.2.1.

Alternatively, read the MkRules file and set the macros as needed for your 
compiler (as the manual suggested).

However, gcc 3.4.2 is ancient and has known bugs that means it cannot even 
compile R correctly on Windows.  If you must use gcc 3, at least use 
3.4.5.


-- 
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