On Tue, Jul 18, 2017 at 8:04 AM, Tom Hughes <tom(a)compton.nu> wrote:
I'm not necessarily concerned about subverting (or
"changing" as I would
prefer to think of it) the One True Fedora Way, if there are good reasons to
do so and it improves things.
I don't know what you mean by One True Fedora Way. Sounds like mythology to me.
By which I mean that there will no longer be one way to update a
system because workstations and servers have to be managed in completely
No, with modularity and layering, my expectation is all the products
will be managed the same in this regard.
In other words that it's no longer possible to use clusterssh to run "dnf
update" on twenty machines at once because each one is now a special
snowflake that requires it's own approach to updating things - possibly even
a mix of approaches if a developer's workstation needs "server updates" to
update postgres and apache and "workstation updates" to update gcc and
There will be a higher level updating command that will permit you to
do updates across all the layers in one whack, as you describe here.
And there will be probably more than one way for admins to do OS
updates requiring reboots independently of application updates which
don't require reboots. If anything you will have more granular
control, rather than less. But there will also be much greater
consistency among those updated machines.