On Tue, 2015-09-08 at 14:02 -0600, Chris Murphy wrote:
On Tue, Sep 8, 2015 at 11:28 AM, Owen Taylor
<otaylor(a)redhat.com>
wrote:
> And on the other hand, it doesn't provide the
> biggest advantage that we can offer to users: the assurance that
> we've
> actually tested not just individual packages that we're installing
> on
> the system, but the actual same operating system that they are
> running.
Why is this not adequately solvable with the exist repo system by
adding, e.g. "validated" and "validated-testing" repos? Then people
who want the old way still use fedora+updates, testers additional use
updates-testing; and those who want the new way use fedora+validated
and testers use validated-testing?
Anything done by simply putting different things in repos can only
address testing of individual packages, not the combination of packages
on a system. There's more discussion in
https://lists.fedoraproject.org/pipermail/desktop/2015-July/012567.html
and the linked-to Wiki page.
One way I like to think of it, is that I can do all sorts of things to
my Fedora system
- without changing package content
- without breaking 'rpm -Va'
That make it misbehave in minor or major ways. As an OS developer,
that's great flexiblity. As a user - that's not so good - because if
any of those things happen to your system by chance or by poorly tested
upgrade paths, then your system will never recover on its own.
- Owen