Yeah, let's add that automation to the post-0.5.0 activities...
From: John Peterson [mailto:peterson@...]
Sent: Thursday, June 09, 2005 12:12 PM
To: KIRK, BENJAMIN (JSC-EG) (NASA)
Subject: RE: [Libmesh-devel] 0.5.0 code freeze
KIRK, BENJAMIN (JSC-EG) (NASA) writes:
> Portability update:
> I've checked:
> - gcc 3.4.3, 4.0.0
> - icc v7.1, ia32
> - icc v8.0, ia64
> - IBM xlC (longhorn)
> - gcc 3.3, OSX
> - gcc 3.3.3, cygwin
> So far, so good. Only two changes:
> - HAVE_PETSC test in ex8.
> - Nonlinear solver support for petsc-2.1.1
> Note that gcc-2.9x is being silently dropped. That is, I'd rather not >
advertise it as supported, but with minor changes one could make it work. >
The final straw is that the std::distance implementation in 2.95.3 is not >
right. Instead of >
> distance_type std::distance (Iter &begin, Iter &end);
> they have
> void std::distance (Iter &begin, Iter &end, distance_type &d); >
> A viable option for someone stuck with 2.9X would be to install STLPort
and > use it instead of the compiler-provided STL implementation.
That's fine with me, there was a bunch of "using" declaration stuff that
didn't work right with gcc-2.9X either so I'm happy to get rid of it.
Would it be possible for you (you have access to the most machines I think)
to automate this process at all, like a nightly builds thing? Wolfgang uses
a nightly build script with cron jobs I think on the machines where he has
accounts. Could be worth trying.