Problem
getCommentVolumeByDay (docket page) scans the comment partitions in the browser to build a per-docket daily series. Separately, the feed's surge signal can only use the monthly partition delta from comments_index — the finest "recent" grain available today — because feed_summary carries no comment time-series.
Proposal
Bake a per-docket daily comment volume series at ETL, alongside the per-docket cluster summary (see the comment-clustering precompute issue). The doc notes it's cheap to materialize.
What it unblocks
- Docket page reads the series directly instead of scanning partitions
- The surge discovery signal can use a true rolling window instead of monthly deltas
Source
§2 of the front-end compute architecture doc.
Problem
getCommentVolumeByDay(docket page) scans the comment partitions in the browser to build a per-docket daily series. Separately, the feed's surge signal can only use the monthly partition delta fromcomments_index— the finest "recent" grain available today — becausefeed_summarycarries no comment time-series.Proposal
Bake a per-docket daily comment volume series at ETL, alongside the per-docket cluster summary (see the comment-clustering precompute issue). The doc notes it's cheap to materialize.
What it unblocks
Source
§2 of the front-end compute architecture doc.