[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: smart package mgr question?

On Mon, Oct 02, 2006 at 08:14:28PM +0530, Rahul Sundaram wrote:
> Axel Thimm wrote:
> >On Sun, Oct 01, 2006 at 01:14:35AM +0530, Rahul Sundaram wrote:
> > RPM is not designed for downgrades
> >
> >One of the core features of rpm from the very beginning was to
> >downgrades and uninstalls so the user is able to revert from a bad
> >package.
> I didnt see any RPM design documents

Cool, you found rpm design documents?

> mention downgrades so I am curious where you got this impression
> from. A few examples,
> http://fedora.redhat.com/docs/drafts/rpm-guide-en/ch01s02.html
> http://www.rpm.org/max-rpm/s1-intro-to-rpm-rpm-design-goals.html
> While there are command line arguments that would let RPM perform 
> downgrades and even ignore dependencies, I wouldnt claim it was designed 
> for such things.

Rahul, I don't see what you're trying to get at. It is known that both
documents have been written after the fact and that maximum rpm
contains many inaccuracies even though it was the sole documentation
available for users for several years. Still even maximum rpm has this
to say about downgrades (as you would had found yourself with a
little more research):

  "With RPM, there's no need to name the programs with a
   version-specific name, as RPM can easily upgrade to a new version
   and even downgrade back, if the new version doesn't work as well."

But the best and most accurate source is to just grep through rpm's
source for "downgrade", or ask someone else from the early days to
confirm what I've stated if you don't trust me. redhat.com's bugzilla
and the rpm user and devel lists have also references to downgrade
support both in rpm mechanics and explicit packages. Also ask yourself
the question: What good would the (now also ancient) repackage support
be if downgrades were not part of the design? So downgrade support is
at least as old as repackaging support, but in fact is even older,
call it pre-historic if you like. ;)

Anyway it remains a fact that rpm properly handles downgrades just the
same as upgrades, and that is not by coincidence, but by (very early)

Furthermore it's also a fact that if a Red Hat/Fedora package is not
downgradable it is considered a bug. Special cases like
non-downgradablity due to change of semantics (e.g. downgrading the
FC5 kernel to RH7.3's) excluded. Otherwise you would lose an important
aspect of packaging: Being able to revert a change.
Axel.Thimm at ATrpms.net

Attachment: pgpli6ugaStkP.pgp
Description: PGP signature

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]