On Fri, Jul 30, 1999 at 07:55:13PM -0700, Joseph Carter wrote:
> read "mv" as "cp, verify success, rm old, create symlink, and the whole
> time deal with things like dropped .dhelp files in /usr/doc while the rest
> of the package has moved to /usr/share/doc already"
...which of course means if you *do* have /usr and /usr/share on the same
partition, you need to have as much free space as /usr/doc takes up, whereas
with an ordinary mv (rename(3), anyway), you need no such thing.
And doing it the way mv(1) does means failures halfway through leave
you with files in /usr/doc/foo and /usr/share/doc/foo, which could be
hard to deal with correctly.
> Not trivial. But if it were trivial we'd have done it already.
> Fortunately if this script does something like lock the dpkg database
> while it does this (a good idea since we're reading that) it should be
> acceptable.
This probably means such a script shouldn't be run from a postinst
(where the package database is already locked, but not only that, cached
by dpkg).
If it's not run as part of the postinst, we're left with the admin manually
cleaning up after the packaging system, which was what I thought we were
trying to avoid.
FWIW, I'm really uncomfortable with having dpkg's database modified by
hand. Looked at, I can deal with, if only because dpkg is so painfully
slow. Modifying it just seems like a good way to destroy systems in
horrid and fanciful ways.
I'm tempted to object to any such proposal that doesn't have the support
of Ian Jackson or Klee Dienes or someone equally familiar with dpkg
internals.
Cheers,
aj
--
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. PGP encrypted mail preferred.
``The thing is: trying to be too generic is EVIL. It's stupid, it
results in slower code, and it results in more bugs.''
-- Linus Torvalds
Attachment:
pgpphmOWbkVi7.pgp
Description: PGP signature