WARNING:DO NOT UPGRADE TO CORE 4

[Les Mikesell]

On Thu, 2005-07-14 at 08:02, Timothy Murphy wrote:

> However, the only point I am trying to make
> is that when I followed the instructions to make a clean install
> it did not work, for reasons which I have already explained
> (many times).

Agreed: sometimes you have to do things that are not optimal
to work around bugs or other pragmatic issues.  My point is
that you should realize that such workarounds may have other
side effects.

> > The difference 
> > is that now you don't know precisely what else you still
> > have left over from the last system besides the part that
> > luckily works around a new bug.  The effects from the other
> > parts may not be so lucky for you.
> 
> I don't know what you mean.
> I have never had any problems of this kind when upgrading.

The fact that there are differences you don't know about may or
may not ever result in problems.  As you've seen, some things
may even be better.  The point is that it isn't predictable
anymore.  

> The main advantage of upgrading as far as I am concerned
> is that you don't have to worry about re-writing
> CUPS config files, etc.

What do you expect to happen when new versions of programs
installed during an upgrade have new options that would
have values supplied in the config files provided by a
fresh install but that is omitted in your inherited verson?
This possibility is one of the main reasons that RH/Fedora
distributions do not do application version-level updates
within the same OS version lifetime but instead back in
only essential bugfix and security patches trying not to
change expected behavior or config requirements.  Between
OS versions, all such bets are off.  Even worse, there may not
be an exact correspondence between packages from one release
to another so some things that should be replaced might not
be.  That fact that upgrades work at all shows that someone has
put some effort into the process, but I wouldn't trust it to
be a priority to get every little thing right even for the
instances where there is a 'right' way.

> But my original point remains:
> people should not be so dogmatic about what works or does not work,
> based on their experience with one (or even several) machines.

I'm not being dogmatic about not doing things like that to work
around problems.  I'm being dogmatic about understanding that such
workarounds may make the machine unpredictable.  If some other issues
appear, you won't know without testing on other machines whether
the problem was created by something inherited from a prior
install or not. 

> On the other hand, the fact that you have a problem installing FC-4
> (or even three problems, like the OP)
> does not mean that Fedora is a disaster.
> It just means you have to knuckle down and analyse the cause of the failure.

What I like to do is keep a copy of the entire /etc directory along with
/home from the prior install.  Then if there are any problems with the
new version, diff the relevant config files between the two versions to
see what made the old one work.  This would be painful in your case of
a scsi controller mismatch but for most other things you can get far
enough to do that.

-- 
  Les Mikesell
    lesmikesell at gmail.com