Re: [gentoo-dev] baselayout redefines /etc/fstab

[Marius Mauch]

On 02/07/04  Seemant Kulleen wrote:

> On Sat, 2004-02-07 at 20:06, Steven Elling wrote:
> > On Wed, 2004-02-04 at 13:48, purslow@×××××××××.ca wrote:
> > > recently, i re-emerged 'baselayout',
> > > which caused a set of candidates to be created for 'etc-update'.
> > > most were innocuous or easily understood,
> > > but one was for  /etc/fstab , which seems both dangerous &
> > > unjustified.
> >
> > I agree!
> >
> > This has been brought to the attention of the devs already,
> > discussed at length, debated at length, and eventually ignored as if
> > it is not a problem.
>
> Not so.  It has been discussed and debated, yes, but not ignored.  The
> only reason for the perceived inaction is really the absence of an
> acceptable solution.  I'm toying with the notion of creating the .cfg
> file using the user's installed fstab file so that the diff is against
> a locally known quantity rather than a blind default quantity.  I'm
> not sure on the pros and cons and haven't given the thought much air
> time for discussion yet, so I guess consider this the official
> throwing out.

Just thinking about this, how about the following procedure:
- if a config file is installed the first time also let portage make a
backup to a safe place (say /var/db/pkg/cat/pkg-ver/config)
- whenever the file is target to be updated make a diff against the
original config file, replace the backup with the to-be-installed file
and use the diff to create the .cfg* file.
- Optionally ignore comments/cvs headers in the diff

Maybe that's what you meant, just formalizing it a bit more ;)

Marius

-- 
Public Key at http://www.genone.de/info/gpg-key.pub

In the beginning, there was nothing. And God said, 'Let there be
Light.' And there was still nothing, but you could see a bit better.