FC5 T2

[Jeff Spaleta]

On 1/15/06, John Summerfied <debian at herakles.homelinux.org> wrote:
> It seems to me that Red Hat has a management problem it's ignoring.
>
> It needs to address the facts of life and come to terms with them. User
> behaviour is human behaviour, and it will not change at Red Hat's behest.
>
> I think RH will serve itself and its users better by adopting the
> policy, when it's ready it ships.

I say go further.. i say fedora core releases every single release
candidate tree to the public servers into the same directory and then
only after its clear that a particular release candidate tree is what
the release team is expecting.. tag that as the final release.  That
way the mirrors are syncing potentially smaller changes between
release candidates into the release candidate tree and no one really
knows which release candidate respin will end up being the final
release. When it is finally tagged the mirrors have all the bits ready
to go without waiting 3 days for the full sync.  In the meantime
you'll have 6 ot 7 potentially different pre-released isos in the 4 or
5 days leading up to the release... all marked as on disk as fcX...
all with some sort of egregious problem tha necessitated a new
installer re-spin.

The only way you'll be able to know for sure if the isos you got are
official is doing the checksum against the official checksums.  I'm
absolutely sure the first such fedora release that followed this very
open release candidate exposure model would be very painful for a
number of users.  But subsequent releases would see far less
unqualified discussion about leaks isos before release dates... simply
because it will be nearly impossible for anyone other than the release
team to know more than 8 hours ahead of time which release candidate
image will be the final release. If people want to roll the dice and
pick open mirrors knowingly undermining the release model.. lets give
them so much release candidate garbage that its not worth the effort.

-jef