Learning from the future of component repositories

An important aspect of the quality assurance of large component repositories is the logical coherence of component metadata. We argue that it is possible to identify certain classes of such problems by checking relevant properties of the possible future repositories into which the current repository may evolve. In order to make a complete analysis of all possible futures effective however, one needs a way to construct a finite set of representatives of this infinite set of potential futures. We define a class of properties for which this can be done. We illustrate the practical usefulness of the approach with two quality assurance applications: (i) establishing the amount of "forced upgrades" induced by introducing new versions of existing components in a repository, and (ii) identifying outdated components that need to be upgraded in order to ever be installable in the future. For both applications we provide experience reports obtained on the Debian distribution.