Re: intent behind multiple versions of the same package at the same time.

[Jeff Johnson <jbj JBJ ORG>]

On Thu, Mar 15, 2001 at 12:50:16PM -0800, Gregory Leblanc wrote:
> On 15 Mar 2001 12:02:17 -0800, Thornton Prime wrote:
> > On 15 Mar 2001, Gregory Leblanc wrote:
> > > Hi there!  I've been looking at some of the features that RPM provides,
> > > and I'm not quite sure how this particular one was intended to be used.
> > > I'll post some thoughts about this feature in followups, but I wanted to
> > > know how this was being used by other people, and what the intent behind
> > > it was first.
> > 
> > It is all but necessary for the kernel. You should never upgrade over an
> > existing kernel. You should always install the new version, reboot, make
> > sure things work for a few days, then go back and remove the old version.
> 
> How can you ensure a clean upgrade path for packages when you do this?
> For the kernel, it's clearly not an issue, but for packages which you
> might want to upgrade, it seems messy.  For example, let's say I have
> libfoo-1.0.1-2, and libfoo-2.0.0-1 installed.  It turns out that
> libfoo-1.0.1 has a security hole, and they release libfoo-1.0.2-1.  How
> does an upgrade handle this situation?  If it doesn't, then I guess
> multiple versions of the same package should only be used for things
> that don't -ever- get replaced, right?
> 

Depends on what you mean by "clean upgrade path". It's certainly pssible to
define and implement upgrade behavior for rpm in the face of multiple packages
already installed, some "older", some "newer", etc.

That said, aside from multiple kernel's installed, there's little reason
to have multiple copies of the same package installed, you'll usually
just get confused.

Now if I can just get gnorpm to remove the INSTALL button from their GUI ...

73 de Jeff

-- 
Jeff Johnson	ARS N3NPQ
jbj@jbj.org	(jbj@redhat.com)
Chapel Hill, NC