My vote (if it matters) is to re-evaluate the "incredible shrinking" toolkit. The conservative approach would be to un-obsolete the initial package. If we continue in the same direction, the only safe package is the core, because it doesn't depend on other applications and services.
I think this is a *very* bad trend.
Do you pull the loose thread until the garment no longer exists? Or, do you repair the loose thread and retain the garment? One way you end up with a patchwork, but at least you don't end up standing in public with no clothing.
In cases where someone decides to "obsolete" some toolkit functionality and it's later found that this decision wicks to other pakages, it's time to re-evaluate that initial decision. At the very least, this situation proves that there was inadequate research and discussion that went into the initial decision.
