v134 · miscellaneous

Private Aggregation API: per-context contribution limits for Shared Storage callers

Enables Shared Storage callers to customize the number of contributions per Private Aggregation report.

concepts

  1. PA Context Limits

    Per-context contribution limits for Private Aggregation. Fairer aggregation budgets across multiple worklet callers.

  2. Budget tuning

    The two motivating cases: callers that want more than 20 contributions vs. callers paying for unneeded noise padding. Dial the cap and watch the assembled report.

  3. Noise Padding Visualizer

    Cell-by-cell visualisation of the report layout. Real contributions blue, padding grey, dropped red. Pre-v134 (fixed 20) vs v134 (caller-chosen) side by side.

  4. Multi-Caller Planner

    Plan contribution limits for multiple Shared Storage callers in one page context. Add callers, set per-context limits, see the report assembly, spot over-budget callers, and generate the privateAggregation.setContributionLimit() snippet for each.

why it shipped

Private Aggregation imposes a limit on the number of histogram contributions that can be embedded in a single aggregatable report, with any additional contributions being dropped. Shared Storage callers can technically work around the limit by invoking another Shared Storage operation, but it's not ideal. Some callers may even want fewer than 20 contributions per report, but for privacy reasons all reports are padded up to the fixed size.

references