[R] mysterious error on compile R 2.3.1

[Prof Brian Ripley]

On Wed, 20 Sep 2006, Dylan Beaudette wrote:

> Getting a very strange error with a new install of R from source on x86;
>
> make[3]: Leaving directory `/tmp/R.INSTALL.r20887/cluster/src'
> ** R
> ** data
> **  moving datasets to lazyload DB
> Error in factor(c(1, 2, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1),  :
>        invalid labels; length 2 should be 1 or 1
> Execution halted
> ERROR: lazydata failed for package 'cluster'
> ** Removing '/home/dylan/src/R-2.3.1/library/cluster'
> make[2]: *** [cluster.ts] Error 1
> make[2]: Leaving directory `/home/dylan/src/R-2.3.1/src/library/Recommended'
> make[1]: *** [recommended-packages] Error 2
> make[1]: Leaving directory `/home/dylan/src/R-2.3.1/src/library/Recommended'
> make: *** [stamp-recommended] Error 2
>
> note that i am using the GCC flags:
> CFLAGS=-march=opteron -ffast-math
> CXXFLAGS=-march=opteron -ffast-math

We do request you do not do so in the R-admin manual.  From the gcc man 
page:

        This option should never be turned on by any -O option since it can
        result in incorrect output for programs which depend on an exact
        implementation of IEEE or ISO rules/specifications for math func-
        tions.

R is such a program.

-- 
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