2010/2/17 Patrick O'Callaghan <pocallaghan(a)gmail.com>:
On Wed, 2010-02-17 at 15:49 +0100, Thomas Janssen wrote:
> > 2) not restricting yourself to kde* (and its dependencies) increases
> the
> > risk of a flaky system because you're updating a bunch of stuff you
> > don't care about. It can then be hard to back out to *not* using
> > updates-testing for every future update.
>
> Well. Not true. Whatever is in updates-testing, you should care of.
> Most updates are bugfixes and come in 99.99% of the cases into updates
> anyways.
I mean that unless I'm prepared to do testing and reporting, I don't
care about them because they'll eventually appear in updates.
It's about the risk of a "flaky" system you said. I use
updates-testing and kde-redhat since a long time. And i had some small
problems with it. Everything easily fixable trough a downgrade in the
worst case. And if you face a problem that hit not every user but the
package is already in stable you have to be prepared to report the
problem anyways.
> And you can easily back out by disabling the updates-testing
repo.
"Backing out" means reversing an earlier advance, e.g. by using "yum
downgrade". Disabling updates-testing won't do that. All it does is
tread water until the standard repo catches up. Nothing wrong with that,
but it's not "backing out".
Ok, so backing out means to revert the update trough a yum downgrade.
Where's the problem?
"Oh, i'm not convinced with the packages, so i disable updates-testing
and run yum downgrade". In the worst case grab the packages from a
mirror and run: rpm -Uhv --oldpackage
> Nothing brakes, no problem with future updates. We're not
speaking of
> enabling rawhide for some updates. That will give you headaches at
> some point.
"breaks", not "brakes", but fair enough.
Yeah, my bad. But glad you spotted my typo. Though the truth.
--
LG Thomas
Dubium sapientiae initium