On Thu, 2012-01-19 at 15:30 -0800, Adam Williamson wrote:
On Sat, 2012-01-14 at 19:12 +0100, Kevin Kofler wrote:
> Kevin Fenzi wrote:
> > Keeping packages around with no maintainers or people handling their
> > bugs is poor for everyone.
>
> Why? If I, as a user, really need a certain piece of software, I'd rather
> have an unmaintained package than none at all! Worst case, I can't use the
> package at all, in which case I'm still no worse off than with no package at
> all!
I disagree. The existence of a package triggers certain assumptions: the
package will be maintained and keep working. That's the point of there
*being* a package, after all. So if there's a package for something, I
don't check for security updates for that 'something' myself. I figure
the packager is doing that for me.
So if I wind up with an unmaintained package installed, my security has
just been reduced.
Yes, I agree with this completely. If something is not being maintained
in Fedora, it's better to retire it. Then a user who wants that piece of
software will have two options:
1) They can build it and maintain it themselves on their own system(s)
2) They can choose to build and maintain it for Fedora by unretiring it.
Either way, they will not be given a false sense that the package is
being maintained.
> (And now with my packager hat on, fixing and/or updating a
package in
> the repo also requires less effort than unretiring a package which got
> removed.)
This is an important point: I think it would be much less of a problem
to retire packages if the process for unretiring them were not so
painful. I _do_ think the unretiring process is an excellent example of
unnecessary bureaucracy (as is the renaming process, good lord, what a
PITA). Those two things could stand to be trimmed down. At least to 'if
you're a provenpackager (or even just a sponsored packager) you can
unretire a package without any obstacles'.
If you file a FESCo ticket to change this policy, this approach would
have my support. There's no reason that a package rename or unretirement
should need to go through a full review (although as I said in an
earlier email, the side-effect here is that such things can help curb
specrot).