As with lsst.verify, we worry that there is confusion about how to distinguish metric values measured on different versions of the codebase, configurations, datasets, etc within the
SQuaSH system. For example, it is possible to define and track a metric on an algorithm implemented by code in the lsst.pipe.tasks package. But that code may be run in multiple
different contexts: as part of alert production, data release production, recovery, etc: how
does SQuaSH distinguish between all of these environments? We believe that SQuaSH is capable of this, but the existing “big picture” documentation is lacking and hard to follow.