On Tue, 9 Dec 2003, Thomas Dzubin wrote:

> (if a VAX cross-compile on a Intel x86 (Pentium-whatever) works, but
> the same compile on a true hardware VAX fails...could this troubleshooting
>  be helped by running SIMH/vax on the Intel box if the compile speed is
> fast enough?)

It's only been within the past few days that I've managed to get a
cross-compiled -current build running on a VAX.  Most of the problems
I've had since then are with getting my cross-compile host to build a
second iteration cleanly.  I think Matt Thomas may have removed the
last hurdle for that a little earilier today.  I'm not sure that there
will be problems with the native build on a VAX running -current.

My present strategy is basically to hammer away at the cross-compiles
until I can build a release from start to finish without a failure. This
approached seems to have caught a few problems already, and it allows
for quick turn-arounds on the builds.  (And quick feedback and fixes).

Once the cross-compile can be done cleanly, I'll take those sources and
let them compile natively on a VAX over a period of several days.  I'm
not sure yet whether or not there'll be problems compiling -current with
the snapshot I created over the weekend, and it might be another week
before do I know (assuming my latest cross-compile doesn't hit any snags
tonight.)

So, yes, being able to compile -current natively, even under SIMH/vax,
as fast as possible would definitely help with the testing.  The sooner
problems can be found, the sooner they can be fixed!

-brian.