Repository navigation
Dashboard: hen-day lay rate shows over 100% (108%) #943
Description
Observed
On the Dashboard's Lay rate card, the hen-day figure reads 108.3% on main and 107.5% on PR #940's branch. Both come from the sim fixture's default farm (101 flocks, all days partial), in the #940 before/after captures (1280 dark, "orders and Lay rate" frame). A hen lays at most one egg a day, so a hen-day rate above 100% cannot be correct.
It was already present before #940, which only changed the window it is computed over.
Where to look
- The rate comes from the production report (
GET /reports/production, spec §19.3). SeeIReportQueries.cs:HenDayPct = RatedEggs / RecordedHenDays. Its comments describe numerator/denominator matching rules: "the eggs of exactly the flocks whose birds are inRecordedHenDays". - Candidates:
- (a) the numerator and denominator cover different flocks or days (for example, partial days);
- (b) bird counts in the fixture are lower than the eggs recorded;
- (c) the Dashboard's window join over two report calls double-counts.
Done when
- The cause is found and stated: a calculation defect, or unrealistic fixture data.
- A calculation defect is fixed, with a test using literal values, including a partial-day case.
- If the cause is the fixture, the simulation seed is corrected so the Dashboard shows a plausible rate, and the report either caps or flags an impossible rate instead of displaying it silently.
Activity
mforce commentedon Sep 26, 2026
Cause, amended after Codex gpt-6-sol review round 1 of #959: the figures in this issue come from fixture data (Sim House B recorded 380+ eggs a day from at most 350 birds; demo House 1 430+ from 420), but there was also a calculation gap. The exposure walk applied every bird movement dated D after counting D's birds, so a same-day addition (a negative Adjustment correcting a miscount) was missing from D's hen-days: 10 birds plus 10 added on D laying 20 eggs rated 200%. Fixed in 7ac891c on #959: additions count toward the day they are dated, removals keep taking effect the next day, and the over-100 flag decides from the unrounded halves.
mforce commentedon Sep 26, 2026
Cause, final after Codex gpt-6-sol review round 2 of #959: the figures in this issue come from fixture data (Sim House B recorded 380+ eggs a day from at most 350 birds; demo House 1 430+ from 420). The calculation itself keeps the #92 convention: every bird movement, addition or removal, takes effect from the next day. A same-day addition can therefore legitimately read above 100% for that day, and the over-100 flag on Reports and the Dashboard surfaces it instead of capping the figure. A split rule counting same-day additions on the day was tried in round 1 and reverted by owner decision after round 2 showed it inflated exposure on a day whose mortality was corrected and double-counted paired same-day transfers between flocks. Regression tests pin all four scenarios.