Re: Bug#206928: LSB vs. Policy

[Martin Godisch <martin@godisch.de>]

reassign 206928 debian-policy
thanks

On Sun, Aug 24, 2003 at 10:29:55 +0200, Eduard Bloch wrote:

> > > See subject. The init script should just be quiet, as other init
> > > scripts do. Instead, it says "...program not found".
> > 
> > This is in compliance with the Linux Standard Base:
> > 
> >   In case of an error, while processing any init script action except
> >   for "status", the init script must print an error message and return
> >   one of the following non-zero exit status codes. 
> > 
> >   5 - program is not installed
> > 
> > http://www.linuxbase.org/spec/refspecs/LSB_1.3.0/gLSB/gLSB/iniscrptact.html
> 
> I do not care much about LSB when it specifies bull^h^h^h^h weird things -
> and in this case, it clashes with our Policy (§9.3.2).
> 
> | Therefore, you should include a test statement at the top of the script, like this:
> | 
> |      test -f program-executed-later-in-script || exit 0
> 
> "Top" is the second line of effective code. If you think different,
> convince me with a consens.

The Debian Policy says:

  These scripts should not fail obscurely when the configuration files
  remain but the package has been removed, as configuration files remain
  on the system after the package has been removed.

nullmailer's init script does not fail obscureley, it clearly states why.
The policy recommends ("should include...") handling this case quietly. As
long as this is only a recommendation, while the LSB requires a failure
("must print an error message"), I'll go with the LSB. This decision was
made after a lengthy discussion with another bug submitter, and I see no
reason to change it.

I'm reassigning this bug to debian-policy. Feel free to clear up this
Policy vs. LSB issue there.

Kind regards,

Martin