On Mon, 2013-11-25 at 09:23 -0800, Adam Williamson wrote:
On Mon, 2013-11-25 at 13:59 +0100, Christian Fredrik Kalager
Schaller
wrote:
> Hi everyone,
> First of all apologize for this taking so long, I ended up traveling
> non-stop for some time visiting some of Red Hats desktop customers.
> While not directly tied to the work of this working group I do hope to
> take some of the lessons learned from those meetings with me into the
> future work of the working group.
>
> Anyway I tried editing the PRD a bit based on the feedback we got on the
> first draft. I tried to make a few items a bit clearer and also to
> include spelling fixes contributed and so on.
>
> We probably want to do another WG meeting soon to discuss next steps.
>
> Feel free to let me know if I forgot to include some important feedback
> or if further clarifications are needed.
"Upgrading the system multiple times through the upgrade process should
give a result that is the same as an original install of Fedora
Workstation."
Based on my experience (>10 years of it, with multiple distributions and
OSes), this is an incredibly ambitious goal. It may in fact be entirely
unachievable as written. I'm not aware of a single operating system in
existence which actually achieves this. Even cellphone manufacturers -
who have a very clearly-defined single piece of hardware to deal with,
and a much smaller set of software and use cases to worry about than we
have - don't achieve this. I'm really not sure it should be front and
centre in a foundational document without some really convincing
evidence that it's even vaguely achievable.
Yes, this is an ambitious goal. I hope we can have ambitious goals for
Fedora workstation. But it's also a really important goal. Currently we
put Fedora users into an impossible situation:
* Fedora releases frequently
* Fedora has a short supported release lifetime
* Every upgrade of a Fedora system is somewhat hazardous
* If you serially upgrade a Fedora system many times, even if there
is no out-right breakage, there is degradation.
The main target of Fedora workstation is a technical user of some sort,
but we can't just assume that they'll know how to fix their system or
have an inclination to do so - most technical users are not operating
system engineers.
If we don't want to support Fedora workstation releases for the lifetime
of the user's hardware (5-7 years), then we need to figure out how to
make upgrades non-events.
An image-based approach to operating system installation and upgrades is
an efficient technical means to this end - but not the only way to get
there. The starting point is a system definition - if any possible
combination of packages from the Fedora package universe with any
arbitrarily changed set of config files is a valid Fedora workstation
configuration, then upgrading can never work.
- Owen