On Wed, 10 Nov 2004, Marc Schwartz wrote: > On Wed, 2004-11-10 at 08:05 +0000, Prof Brian Ripley wrote: > > [Moved to R-devel.] > > > Second problem: gcc 3.4.2 does not work `just fine': it has a pesky bug > > that causes several CRAN packages to fail/give incorrect answers. Now > > Linux distros have a very nasty habit of distributing bug-enhanced > > versions of major software, so this may be fixed in FC3 but in any case > > gcc-3.4.3 is now out and a worthwhile upgrade. (Apparently Mandrake > > backported the bug to `3.4.1' ....) > > > > I would recommend either installing from the sources or building from the > > SRPM. Hopefully by the end of the week we will have some i686 machines on > > FC3 (only x86_64 so far) and so be able to make RPMs available. > > I am finally up and running on FC3 this morning. Took a while to > download the DVD ISO via Bittorrent... > > Using: > > # gcc --version > gcc (GCC) 3.4.2 20041017 (Red Hat 3.4.2-6.fc3) > > I installed: > > Version 2.0.1 beta (2004-11-10) > > from source this morning. > > It passes make check-all without problems. > > Prof. Ripley, is there a particular CRAN package upon which I can test > the bug issues that you refer to above to confirm their presence or > absence on FC3?
I know minpack.lm and nlmeODE used to fail R CMD check in the examples on the real gcc 3.4.2. I've just tested they on an AMD machine under FC3, and they seem OK -- what I am not sure is if they ever failed there (they did on i686). At a quick glance the embedded reference in etc/Makeconf to gcc 3.3.3 when compiling packages with Fortran is not causing problems as 3.4.2 has reorganized where things are. My tests are on a server in a server room, and at present Xvfb won't run on it (I think it needs cleaning up after the upgrade), which limits the testing I can do tonight. Brian -- Brian D. Ripley, [EMAIL PROTECTED] 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 ______________________________________________ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
