[1619] in Release_7.7_team
Re: 8.2.17 compiler workaround
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Thu Jan 28 03:02:59 1999
From: Jonathon Weiss <jweiss@MIT.EDU>
To: Greg Hudson <ghudson@MIT.EDU>
Cc: release-team@MIT.EDU
In-Reply-To: Your message of "Mon, 25 Jan 1999 16:52:53 EST."
<199901252152.QAA21148@small-gods.mit.edu>
Date: Thu, 28 Jan 1999 03:02:51 EST
> The following little script makes the minimal changes necessary to
> make the compiler work again on non-updated machines. It just makes
> the libld symlink and updates the libelf shared library. I've tested
> it.
A couple of comments:
It's itempotent, this is good.
It would probably be a good idea for the script to make suer you're
running something between 8.2.0 and 8.2.15 inclusive, and print a
useful error message if you aren't. Although this probably isn't too
big a deal since things ought to fail gracefully, unless you take some
of my suggestions below.
Is it possible to arrange it so that people can run the script before
8.2.17 goes public, so that they can deal preemptively, and have no
down time. There are a couple of issues with this: 1) people end up
running old binaries off the packs against new libraries (the reverse
skew) and 2) it would probably require getting the libraries out of
the dev cell. I don't feel too bad about copying things off of these
packs, but I don't really like the idea of linking to them explicitly
which brings up my next comment...
Is there a reason you are linking some things off onto the packs
rather than just bringing new copies local?
> I figure we can drop this script somewhere and announce its presence
> in the second mail to release-announce. If no one else has a
> preference for where to put it, it will probably stay in
> /afs/dev.mit.edu/project/release/unscrew-8.2.17.sh, where it lives
> now.
I'd prefer that "add release; unscrew-8.2.17.sh" work, mostly because
it seems more intuitive.
Jonathon
