Jeff Spaleta wrote:
On Tue, Dec 9, 2008 at 10:47 PM, Kevin Kofler <kevin.kofler@xxxxxxxxx> wrote:
How's that different from updates-testing?
It's update-testing with time based flushes to stable with an
infrequent flush rate instead of the stochastic flushing to stable
that we do now.
But the bulk of what he wants is for older stable update to continue
to available to be able to downgrade to if a stable update becomes
'known bad' at some point. Instead of pushing yet another update for
'known bad' updates, he wants to back out to a previous 'known good'
update..assuming it isn't marked as 'known bad' as well. I feel there
is a logical fallacy lurking here concerning how things are classified
as 'known bad' but I'm too tired right now to try to articulate it.
Actually I'd leave it up to the packager as to how to best fix the bug -
and would usually expect it to be yet another update with numbering
going forward with either a fix or a backout of some problematic change.
There would just have to be a way to mark the 'fixed' version to be
installed without the delay, bypassing the one it replaces.
That's sort of optimistic, but at least it gives a second chance to fix
things before they hit machines that you can't afford to break. And a
user would have a choice of which machine he would use to test on.
--
Les Mikesell
lesmikesell@xxxxxxxxx
--
fedora-devel-list mailing list
fedora-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-devel-list
