init.d script removal

[Terry Chan]

This topic has been discussed before.

No, lrm, does not REMOVE everything that lin installs, specifically
because we know sysadmins will have to configure/localize certain
items in the /etc directory.

The Lunar team received no end to the whines, when either a) a module
was updated and it wiped /etc tree entries that were painstakingly
hand configured, or b) other /etc tree entries that were uniquely
generated, like ssl_certs.

To that end the lunar coretools currently do NOT remove most
of the files that belong in the /etc tree.

Your example below of a module that wants to have an /etc/init.d
script with the same name as another module is a different problem
entirely.  The second module is NOT allowed to use the same
name for a binary or a shell script, without at least adding
a CONFLICTS script.  Your best solution would be to rename the
offending /etc/init.d script to a different name entirely.

Terry Chan
------------------------------------------------------
On Thu, Jul 31, 2003 at 01:02:45AM +0200, nestu wrote:
> Hello ladies and gentlemen ;)
> I've been asking a question on irc, that I'd like to 
> put on the list: why doesn't lrm remove init'd scripts.
> I don't know if to call it a bug or not, but with the
> naming convention of calling the script by it use and
> not by the module's name it all mucks up. For example: I 
> prepare a DNS server module, that has an init.d script
> called named. Say the user lrm's the module and installs
> bind. It would b0rk the whole thing up, wouldn't it?
> So policy could be: remove init.d scripts in POST_REMOVE,
> or just leave it there. What do you think? I've choosed 
> the first option for now. Mmmmm, maybe in CONFIGURE a:
> 
> if [ -e /etc/init.d/$MySript]; then
> 	query "Do you want to update $MyScript" y
> 
> and so on... ????
> Thanks for listening,
> C U,
> nestu ;))))