[2448] in SIPB_Linux_Development
AFS
daemon@ATHENA.MIT.EDU (Derek Atkins)
Tue Feb 2 11:34:32 1999
To: linux-dev@MIT.EDU
From: Derek Atkins <warlord@MIT.EDU>
Date: 02 Feb 1999 11:33:47 -0500
I noticed that Sal built "AFS packages" for 2.0.36 -- really, it looks
like he just renamed the modules and built a new package. This is
certainly the Wrong Thing (TM). Bad Sal! No biscuit.
If the 2.0.35 kernel module works for 2.0.36, the right thing to do
would be SYMLINK the new names to the old modules. So, we'd have:
-rw-r--r-- 1 root root 330812 Sep 18 17:54 libafs.2.0.35.o
-rw-r--r-- 1 root root 16 Sep 18 17:54 libafs.2.0.36.o -> libafs.2.0.35.o
-rw-r--r-- 1 root root 333360 Sep 18 17:55 libafs.smp.2.0.35.o
-rw-r--r-- 1 root root 19 Sep 18 17:55 libafs.smp.2.0.36.o -> libafs.smp.2.0.35.o
Given Sal's modules, people might think that libafs was actually built
against 2.0.36, which it was not. This is confusing and just plain
wrong.
BAD SAL. Consider yourself thwapped, so I wont have to do it in
person next time I see you.
-derek
PS: I can either rebuilt the package with symlinks, or I can rebuild
the kernel module against 2.0.36. I don't care which. But the
current situation is just plain wrong. Ideally, I'd like to
incorporate CMU's package to lookup the afs module based on the __iget
symbol version. But that will have to wait.
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL N1NWH
warlord@MIT.EDU PGP key available
