hey john, stephen, On Tue, Aug 15, 2006 at 12:17:05AM +0100, Stephen Gran wrote: > The problem is needing a non-essential package during purge. This means > one of two things: dbconfig-common becomes essential (not that great, > IMHO - it's big enough already), or the rest of the world that relies on > it uses clever postrm snippets that do the right thing when it's > unavailable. That sucks as well, since it means reinventing the wheel.
i think we'll have to go for the latter. there was a similar problem
a while back when there was stuff i needed to do in the prerm script,
which at the time i was able to avoid through other means.
anyway, the kludge shouldn't be too bad from the maintainer script point
of view. i think something like:
if [ -f /usr/share/dbconfig-common/dpkg/postrm ]; then
# set variables...
dbc_go packagename $@
fi
should do the trick. this problem gets a little more complicated
within dbc itself (needing to check for ucf and not blindly using
other commands), but i think that's all the maintainer scripts
should need to do. or, maybe i'm a bit too tired and missing something?
> Admittedly, this is a policy violation that bacula doesn't have compliant
> postrm scripts, but I'm not sure that it's a problem with dbconfig-common.
yes, bacula is not policy complaint as reported in the bug logs, but
i'll take the blame for having instructed people to do so in the first
place :)
so i think in this case the bug should be cloned with one assigned to
each package, and the bacula one can report being blocked by dbconfig's
until i come up with $solution.
sean
signature.asc
Description: Digital signature
