Malcolm,
On 2006 Jun 16 , at 17.02, Malcolm J. Currie wrote:
> More important were the predominent Linux flavours, and MacOSX.
There are (as you know) two variables here, OS version and platform.
Platform: Any new machines anyone purchases will be Intel. The
compilers and linkers are the same ones, though with different back-
ends, obviously, and I've got the impression there are a few gotchas
there -- we all know how finicky the OS X linker is. Since the
project only has a PPC machine and isn't about to spring for a new
one, this question is of only theoretical interest.
Version: pluto is version 10.3.9 -- that is, fully patched, but one
major release old. 10.4 is current. 10.5 is reportedly due out
about the end of this calendar year, very roughly. The default
compiler on pluto is gcc 3.3, whereas 10.4's default is gcc 4.0.1. I
think the decision is more dependent on what Fortran compilers are
available, and I get the impression there's still a bit of a fuss
here for 10.4. I can't speak to this, because I've still not got
around to putting a Fortran compiler on this machine since I put 10.4
on it (thus I haven't build the USSC for some months; SUN/248 is
getting increasingly out of date).
My own conclusion is therefore that pluto is fine as it is for now:
there's no money to worry about the platform, and no urgency to
change the OS version.
If money miraculously condenses out of thin air by the end of the
year, then a pluto replacement will be 10.5/intel and pluto can
usefully be left as 10.3/ppc; if not, 10.5 will doubtless still
support PPC machines, so that might be a good time to scrounge ~£100
for a 10.5 licence to make pluto 10.5/ppc.
How does that sound?
Norman
--
------------------------------------------------------------------------
----
Norman Gray / http://nxg.me.uk
eurovotech.org / University of Leicester, UK
|