Files
inter-hub/Application/Helper/Controller.hs
Bernd Worsch 878d2577ae
Some checks failed
Test / test (push) Has been cancelled
feat(P4): IHF Phase 4 complete — Outcome Observation and Antifragility Loop
Closes the IHF improvement loop. Full antifragility chain now traversable:
Widget → Annotation → Candidate → Requirement → Decision → Deployment → OutcomeSignal

New artifacts:
- DeploymentRecord (immutable, links DecisionRecord to a deployed version)
- OutcomeSignal (append-only; DB trigger prevents UPDATE/DELETE)
- ChangeEvaluation (one-per-deployment; UNIQUE constraint; 1–5 score)

New capabilities:
- DeploymentRecordsController (index, show, new, create)
- RecordOutcomeSignalAction — capture improved/regressed/neutral/inconclusive signals
- Pre/post comparison panel on deployment show (±30-day event/annotation counts)
- Regression detection — improved signal followed by high/critical annotation
- ChangeEvaluation — idempotent score+rationale per deployment
- Recurrence tracking — cycle count per widget, leaderboard
- AntifragilityDashboardAction (autoRefresh, 5 panels) per hub
- Phase 4 integration tests (T01–T08 logic coverage)
- docs/phase4-summary.md; SCOPE.md updated to Phase 4 complete

State Hub: workstream 07e9c860 → completed

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-29 12:27:30 +00:00

71 lines
3.0 KiB
Haskell

module Application.Helper.Controller where
import IHP.ControllerPrelude
import Generated.Types
import Data.Time.Clock (addUTCTime)
import Data.List (sortBy)
-- Here you can add functions which are available in all your controllers
-- | Returns the set of widget IDs that are currently in regression.
--
-- A regression is defined as: a widget that has an OutcomeSignal(improved)
-- for any deployment, followed by a new Annotation(severity IN high/critical)
-- created more than 1 day after the signal's observed_at (grace period).
regressedWidgetIds :: [OutcomeSignal] -> [Annotation] -> [Id Widget]
regressedWidgetIds signals annotations =
[ wid
| wid <- nub (map (.widgetId) signals)
, isInRegression signals annotations wid
]
isInRegression :: [OutcomeSignal] -> [Annotation] -> Id Widget -> Bool
isInRegression signals annotations wid =
let improvedSignals = filter (\s -> s.widgetId == wid && s.signalType == "improved") signals
highAnns = filter (\a -> a.widgetId == wid
&& a.severity `elem` ["high", "critical"]
&& isNothing a.retractedAt) annotations
graceEnd sig = addUTCTime (24 * 3600) sig.observedAt
in any (\sig -> any (\ann -> ann.createdAt > graceEnd sig) highAnns) improvedSignals
-- | Computes the number of completed improvement cycles per widget.
--
-- A cycle is counted when:
-- 1. A RequirementCandidate for the widget was accepted
-- 2. A DecisionRecord exists for that requirement/candidate
-- 3. A DeploymentRecord exists for that decision
-- 4. A new RequirementCandidate was subsequently created for the same widget
--
-- Returns a list of (widgetId, cycleCount) for widgets with cycleCount >= 1,
-- sorted descending by cycleCount.
widgetCycleCounts
:: [RequirementCandidate]
-> [Requirement]
-> [DecisionRecord]
-> [DeploymentRecord]
-> [(Id Widget, Int)]
widgetCycleCounts candidates requirements decisions deployments =
sortBy (\(_, a) (_, b) -> compare b a)
[ (wid, cycleCount wid)
| wid <- nub (map (.sourceWidgetId) candidates)
, cycleCount wid >= 1
]
where
-- A completed cycle: accepted candidate → requirement → decision → deployment
completedCycleDeploymentTimes wid =
[ dr.deployedAt
| c <- filter (\x -> x.sourceWidgetId == wid && x.status == "accepted") candidates
, req <- filter (\x -> x.sourceCandidateId == c.id) requirements
, dec <- filter (\x -> x.requirementId == Just req.id) decisions
, dr <- filter (\x -> x.decisionId == dec.id) deployments
]
cycleCount wid =
let deplTimes = completedCycleDeploymentTimes wid
-- For each completed cycle, check if a subsequent candidate was created
widCandidates = filter (\x -> x.sourceWidgetId == wid) candidates
in length
[ ()
| deplTime <- deplTimes
, any (\c -> c.createdAt > deplTime) widCandidates
]