Re: Autorollback patch question...

[Adam Spiers <adam spiers net>]

Jeff Johnson (jbj@redhat.com) wrote:
> On Mon, Jan 26, 2004 at 06:50:24PM +0000, Adam Spiers wrote:
> > I'm not sure I understand where Jeff was going with his suggestion of
> > actions being based on the result of %post success however.  I had
> > always envisaged the semantics of %post being "try to finish up
> > cleanly by running %post, but if it fails, we consider it an
> > unfortunate partial success rather than a failure which then gets
> > rolled back".  If %post succeeding is to be considered a strict
> > requirement for success, then rpm should be more diligent about
> > cleaning up from its failure.
> 
> Depends on POV. rpm was designed as an installer library, not as
> a general purpose software manager. Within that narrow "installer
> library" design, failures, particularly scriptlet failures, are
> not supposed to happen, that's what package QA is supposed to
> identify and fix long before the installer sees the package.

That's an interesting POV.  In that case I'd say rpm has outgrown its
original design - there are many rpms out in the wild with little or
no QA.  I can see how Red Hat could take the "not our problem" stance
on that though.

> It's really not that hard to check package quality by installing
> in a chroot. It's simply not possible to recover from all possible
> errors, nor even a significant subset of all possible errors.

Agreed - it's not possible to achieve a 100% recovery; however there
is the potential for a best effort recovery, and that grey area is
more or less what we're discussing here.

Finally, I'd always thought of %pre as (partially, at least) an rpm's
chance to declare itself unsuitable for installing via a deliberate
non-zero exit code.  Would you say that's a misplaced belief?