Re: Rationale for /etc/init.d/* being conffiles?
On Fri, Dec 19, 1997 at 01:56:35PM -0500, Scott Ellis wrote:
> On Fri, 19 Dec 1997, Santiago Vila wrote:
[snip policy]
> >
> > Could somebody please explain the rationale for having *all*
> > /etc/init.d/* scripts as conffiles?
[snip]
> You can deactivate OR CHANGE THE BEHAVIOR of the program by modifying the
> script. If it isn't a conffile, this will break every time the package is
> upgraded (I frequently tweak the behavior of the init.d scripts). I know
> I'd get seriously annoyed if my changes were suddenly overridden. The
> only inconvenience there is the occasional prompting about a file I've
> changed being updated. I far prefer that small warning to having programs
> change behavior under my nose. You only got the warning because a package
> was changed to conform to policy and added a conffile that wasn't there.
> For the most part, dpkg does a good job of not bothering you about files
> that you haven't changed or which haven't changed upstream.
What about "dpkg-divert"? Sure - some people do edit /etc/init.d/whatever
(particularly "network"), however there are many files in /etc/init.d that
the vast majority of people won't change. If some behaviour needs to change,
they may not install the new version (I normally say "N" to replace
conffiles and then go through manually to see what's changed). On my
system, there are more than fifty scripts in init.d, I've changed one.
The policy could be changed to remove the *requirement* that they are
conffiles, whilst explaining why it might be a very good idea to make them
conffiles. If the maintainer still doesn't want it to be a conffile, they
could put a note about dpkg-divert at the top of the script.
Adrian
email: adrian.bridgett@poboxes.com | Debian Linux - www.debian.org
http://www.poboxes.com/adrian.bridgett | Because bloated, unstable
PGP key available on public key servers | operating systems are from MS
Reply to: