Previously, aggregation was specified as if all output shares were aggregated into an aggregate share in one fell swoop, at time of collection. However, all implementations instead used an incremental aggregation process to keep a "running total" aggregate share which was updated as aggregation jobs completed, for performance reasons. This commit switches the specification to match the implementations.
Phrasing things this way also makes the name "aggregation job" make sense; as previously specified, there was no aggregation occurring inside of an aggregation job!
Previously, aggregation was specified as if all output shares were aggregated into an aggregate share in one fell swoop, at time of collection. However, all implementations instead used an incremental aggregation process to keep a "running total" aggregate share which was updated as aggregation jobs completed, for performance reasons. This commit switches the specification to match the implementations.
Phrasing things this way also makes the name "aggregation job" make sense; as previously specified, there was no aggregation occurring inside of an aggregation job!