On Sat, Mar 23, 2019 at 2:26 PM Rich Freeman <rich0@gentoo.org> wrote:

> On Sat, Mar 23, 2019 at 10:17 AM Alec Warner <antarus@gentoo.org> wrote:
> >
> >
> > Avoid letting the perfect be the enemy of the good here.
>
> Indeed, we need to avoid treating packages as unmaintained simply
> because they have open bugs.
>
> Many packages have bugs that are fairly trivial in nature, or build
> issues that only show up in fairly obscure configurations.  These
> often affect only a single user.
>

So this is why I advocate for building a number of signals, and using a
combination of signals to determine if a package is unmaintained.


>
> If we treeclean the package we don't actually fix the problem - we
> just drive it to an overlay.  Now instead of a package that works for
> 11/12 users and has an obscure but, we now have a package that isn't
> getting monitored for security issues, and other QA issues that might
> actually be fixed if they were pointed out.
>
> > Rules:
> > A package is unmaintained if it:
> >   - Has not been touched in 5 years
>
> Do we really want to bump packages just for the sake of saying that
> they've been touched?  That seems a bit much.
>

I'm not saying "we should absolutely remove packages that have not been
touched in N years" but I am saying "we should review packages that have
not been touched in N years".


>
> >   - Is behind 3 versions AND hasn't been touched in 2 years
>
> If we have the ability to detect if a package is behind upstream,
> perhaps we should actually file bugs about this so that the maintainer
> is aware.
>
> However, the fact that a newer version exists doesn't necessarily mean
> that there is a problem with the older version.  For some types of
> software a maintainer might be picky about what updates they accept.
> For example, they might need to synchronize versions with other
> distros that update less often/etc.  They should of course accept
> contributions from others willing to test, but the fact that somebody
> is maintaining a package on Gentoo doesn't obligate them to always
> support the latest version of that package.
>

> Now, obviously if there is a security issue/etc then we should follow
> the existing security policies, but those are already established.
>

Would you be happier if there was some kind of opt-out or whitelist?

Have you looked at mgorny's recent removals? its mostly stuff that doesn't
build and hasn't been touched in 5 years and *yeah* I want that stuff out
of the tree; its a net negative for everyone. Keeping packages in the tree
isn't free.


>
> --
> Rich
>
>