Proposed guideline for init script files

[Jesse Keating]

On Tuesday 06 March 2007 12:37:34 Steve Dickson wrote:
> Question: what right do we have to destroy our users initscripts
> when they changed? There has been an expectation for a number of
> years, in a number of packages that if an initscript that has
> changed it will be preserved on updates.. So why should we change that?

For the simple fact that init scripts should _not_ be configuration files, 
_at_ _all_, and the packaging system should not encourage people to treat 
them as such.  Configuration must happen in real config files so that it can 
be preserved while things in the init script like binary name and necessary 
options can change when the application changes.  I personally feel that the 
init script needs to live somewhere outside of /etc to make this even more 
clear, but that's a much larger change that I can't necessarily drive right 
now.  If your customers have to do configuration in the init script, that is 
a bug in your software that needs to be fixed.  If they have to work around 
some bug that we haven't fixed yet, they really should copy the init script 
to a new name and disable the old script, or exclude that package from being 
updated.  We're really trying to clean up the system and consolidate 
configuration into concise areas rather than scattered about the file system.  
These are the changes that Fedora can make moving forward.  If RHEL doesn't 
want to play along, that's their choice.

The Packaging Committee, the Fedora Engineering Steering Committee, and your 
own peers within Red Hat have approved this guideline, and it will be moved 
in hopefully today.

-- 
Jesse Keating
Release Engineer: Fedora
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/fedora-maintainers/attachments/20070306/d56f4b5c/attachment.sig>