[6956] in SIPB bug reports
Re: Digital Unix binaries in sipb locker
daemon@ATHENA.MIT.EDU (Mike Whitson)
Tue Aug 11 18:55:19 1998
To: amu@MIT.EDU (Aaron M. Ucko)
Cc: Mike Whitson <mwhitson@MIT.EDU>, bug-sipb@MIT.EDU
From: Mike Whitson <mwhitson@MIT.EDU>
Date: 11 Aug 1998 18:55:11 -0400
In-Reply-To: amu@MIT.EDU's message of 11 Aug 1998 18:45:03 -0400
amu@MIT.EDU (Aaron M. Ucko) writes:
> Unless you expect your builds to be particularly less reliable than
> the builds for other architectures
The only reason I would expect this is if there are lingering
portability bugs in various sipb-locker code, including
64-bit-cleanliness bugs. I'll try to fix those as I run across them,
but I make no guarantees.
> I'd advocate putting the software in sipb, because
> (1) There's precedence (the HP/UX binaries).
> (2) It produces a more consistent environment with less chance
> of confusion; in particular, you don't have to take any
> special measures (such as making symlinks or specifying
> mixed paths) for shared data files.
> (3) Athenized Alphas' users presumably know to expect less
> support in general.
Okay, I think I concur with Matt's and your reasoning. I'll start
populating sipb/machtype/alpha in the near future. If there are any
"support" questions, feel free to refer them to me.
I'm going to create a separate sipb.alpha AFS volume for the work, but
in the interest of not wasting disk space, I'm not going to bother to
make it replicated unless people feel it's really necessary.
-wrong mike
