User:MGorny/New Treecleaner Policy

Removal criteria
Treecleaning is not an exact science. We do not remove packages as soon as they meet a specific criteria. Instead, we try to determine the state of the package, the scale of the problem, the resulting usefulness of the package and its most likely future.

The facts suggesting removal of a package include:
 * lack of an active Gentoo maintainer (i.e. a maintainer that does not reply to bugs in a reasonable time is not active),
 * lack of an active upstream with an indication that the package is going to require continuous downstream patching,
 * vulnerabilities that are not addressed,
 * major unsolved issues (e.g. build failures),
 * a large number of minor issues,
 * lack of direct usefulness and no reverse dependencies (e.g. unused libraries, Python modules that are not likely useful),
 * blocking a tree-wide migration/cleanup (e.g. using obsolete EAPI, eclass, dependency on a package being cleaned).

The facts leaning towards keeping the package include:
 * willing of the maintainer to fix the bug (in a reasonable time),
 * interest of potential proxied maintainers,
 * a large number of reverse dependencies (might be worthwhile to ping their maintainers to take the package).

Generally, Treecleaners delay planned removals if they prompt action from (potential) maintainers. However, this action should bring measurable results, i.e. it's not acceptable to just say you're going to fix it and not do anything for two months.