[2980] in RedHat Linux List

home help back first fref pref prev next nref lref last post

Re: RPM clobbers files

daemon@ATHENA.MIT.EDU (Donnie Barnes)
Wed Nov 6 11:44:42 1996

To: redhat-list@redhat.com
In-reply-to: Your message of "Wed, 06 Nov 1996 10:07:31 GMT."
             <199611061007.KAA08889@weasel.mmm.co.uk> 
Date: Wed, 06 Nov 1996 11:30:05 -0500
From: Donnie Barnes <djb@redhat.com>
Resent-From: redhat-list@redhat.com
Reply-To: redhat-list@redhat.com

> Donnie Barnes <djb@redhat.com> wrote:
>  [Tim Baverstock <warwick@mmm.co.uk> wrote:]
> >> It sounds like it would be handy to be able to list all those files which
> >> would be clobbered by a given RPM, which no longer match the original
> >> installation.  rpm -Vp, perhaps?  Or have I missed something which does
> >> this already?  (I imagine it wouldn't be too hard to cook something up with
> >> sort and uniq)
> >
> > Come on folks...think about what you are saying.  "All" files that
> > no longer match the original installation would be all the 
> > binaries, libs, etc.  I think here you mean "All" the config files.
> 
> I meant, of course, `the extant installation' - `original' in the sense of
> `before the proposed upgrade' rather than `factory fresh' - that is, any
> file which this RPM intends to clobber, but whose checksum no longer
> matches the one recorded for it in the current RPM database (as per -V).
> I had imagined this to be evident from foregoing posts.
> 
> As someone pointed out, this would usually only flag changed config files,
> but I can think of a particularly nightmarish day when another brand of
> Linux cheerfully deleted a whole slew of customised binaries, only four
> hours after a complete system backup (mercifully). The specifics were
> different, but the principle very much the same.
> 
> I can think of no file which should not be preserved by default if it
> fails to match its signature in the RPM database - even binaries and
> logfiles - with the singular exception of cache files or other transients.

Well, that's just where we disagree.  By your way, if someone hacks your
system, puts their own copy of /bin/bash on the system, and you then upgrade
your copy of bash, the hacked one remains.  It may be nice to *keep* the
file, but *not* leave it in place.

Also, you lose alot of RPM performance.  You'd have to md5 every single file.
Waiting on that to happen on your X server would suck.  Waiting on it for
emacs would suck worse.

> Surely it would be a much safer mistake in RPM building to forget to mark
> something as `can be deleted, even if its signature is different'?
> 
> --saveall could instruct RPM to make .rpmsaves of any files which don't
> match their checksum, even if marked as a transient - a sort of stronger
> default.

We're probably going to implement a --keepconfigs to do that with config
files.

> Granted, we should be vigilant and careful as root, but since the
> computer's much better at spotting this sort of thing than we are, it
> makes sense to me that it should assist, in much the same way that many
> alias rm="rm -i". It's far easier to \rm `find \ -name \*.rpmsave` after
> the event, than back things up beforehand and miss something.

Everything that should *change* should be marked as a config file.  If
you update, you get a .rpmsave on all those.  

> It may also be an idea to have .rpmsave become .rpmsave.YYMMDD.vvvv?
> 
> Forgive me if I sound a bit short, but I'm unused to being told `think
> about what you are saying' in a constructive dialogue, by a relative
> stranger.

Sorry...I'm not running a good track record over the last 24 hours at
being diplomatic.  


--Donnie

--
  Donnie Barnes        http://www.redhat.com/~djb     "Bah."
    djb@redhat.com       http://www.turner.com/lazarusman/   
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
_Things You'd NEVER Expect A Southerner To Say_ by Vic Henley:     
**  I only smoke menthol.



--
  PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
  ________________________________________________________________________
  http://www.redhat.com/RedHat-FAQ   http://www.redhat.com/RedHat-Errata
  http://www.redhat.com/RedHat-Tips  http://www.redhat.com/mailing-lists
  ------------------------------------------------------------------------
To unsubscribe: mail -s unsubscribe redhat-list-request@redhat.com < /dev/null


home help back first fref pref prev next nref lref last post