MCPcopy Create free account

hub / github.com/PerpetualSoftware/pad / functions

Functions28,840 in github.com/PerpetualSoftware/pad

FunctionTestAHeartbeatConsumesNoEventID
TestAHeartbeatConsumesNoEventID is the constraint that keeps the liveness probe from manufacturing the resets it exists to avoid. Three of this bus's
internal/events/redis_heartbeat_test.go:1447
FunctionTestAHeartbeatDoesNotEndAWorkspacesCoverage
--------------------------------------------------------------------------- Phase 1: recognise and ignore. -------------------------------------------
internal/events/redis_heartbeat_test.go:397
FunctionTestAHeartbeatDoesNotEndCoverage
TestAHeartbeatDoesNotEndCoverage is the phase-1 contract, and the half that makes the two-phase roll zero-loss. Fails without the classification: "hb
internal/watchevents/heartbeat_test.go:88
FunctionTestAHeartbeatIsRecognisedByPrefixNotByExactPayload
TestAHeartbeatIsRecognisedByPrefixNotByExactPayload pins the forward compatibility the prefix buys: a LATER version of the frame, carrying fields this
internal/events/redis_heartbeat_test.go:468
FunctionTestAHeartbeatKeepsAQuietWorkspaceOutOfTheDetector
TestAHeartbeatKeepsAQuietWorkspaceOutOfTheDetector is the whole point of publishing them: on a workspace with no events at all, our own frame is what
internal/events/redis_heartbeat_test.go:812
FunctionTestAHeartbeatReachesNoSubscriber
TestAHeartbeatReachesNoSubscriber pins the other half of "bus-internal": a heartbeat must never be delivered as an event, which would put a zero-value
internal/events/redis_heartbeat_test.go:441
FunctionTestAHiddenCrossedBoundChangesNoResponse
(t *testing.T)
internal/store/commitment_links_review_test.go:235
FunctionTestAJoinerIsNotStrandedByAnAbandonedEstablishment
TestAJoinerIsNotStrandedByAnAbandonedEstablishment covers this case. establishSubscription abandons when the workspace emptied while it was dialling.
internal/events/redis_subscribe_confirm_test.go:812
FunctionTestAJoinerIsServedAcrossAnIdleFiredCycle
--------------------------------------------------------------------------- THE JOINT (BUG-2738). The idle detector is a THIRD actor in a region whose
internal/events/redis_heartbeat_test.go:165
FunctionTestAJoinerOfAFailedEstablishmentIsRefusedToo
TestAJoinerOfAFailedEstablishmentIsRefusedToo covers the population the establisher's refusal does not reach on its own: a caller that registered whil
internal/events/redis_subscribe_failed_test.go:216
FunctionTestAJoinersReplayStopsWhereItsChannelStarts
TestAJoinersReplayStopsWhereItsChannelStarts pins the replay ceiling. WHO ACTUALLY NEEDS IT, since the answer is not obvious and I got it wrong once:
internal/events/redis_subscribe_confirm_test.go:472
FunctionTestALapsedBlockerLeavesTheChildBlockedAndALateDoneStillWins
A recorded miss under lapses is lapsed: the child stays blocked for a decision. A Done that arrives late still wins.
internal/store/commitment_gates_outcome_test.go:186
FunctionTestALateRejectionLeavesAdmittedCallersInPlace
TestALateRejectionLeavesAdmittedCallersInPlace pins the residual the fix deliberately leaves: a rejection that arrives AFTER confirmTimeout, once call
internal/events/redis_subscribe_rejected_test.go:170
FunctionTestALegacyWindowsEndEditMovesAnOccupiedUntilEdge
A legacy row derives its roles from kind and dtend, so a dtend edit moves its occupied_until without naming a role.
internal/store/commitment_links_guard_test.go:116
FunctionTestALinkPreviewMovesAndStoresNothing
(t *testing.T)
internal/store/commitment_link_write_test.go:75
FunctionTestALinkThatCrossesTheChildsOwnBoundNeedsAcknowledgement
(t *testing.T)
internal/store/commitment_link_write_test.go:137
FunctionTestALinkToADeadlineIsARenegotiationAtWriteTime
(t *testing.T)
internal/store/commitment_link_write_test.go:90
FunctionTestALinkWhoseOffsetAlreadyMatchesMovesNothing
(t *testing.T)
internal/store/commitment_link_write_test.go:53
FunctionTestALinkWriteBelowALinkedParentRecordsNoDrag
A parent's own incoming link is not a drag: writing a link below it must not record an offset override for the parent.
internal/store/commitment_link_write_test.go:180
FunctionTestALinkWriteLeavesAFinishedChildAlone
(t *testing.T)
internal/store/commitment_links_review_test.go:385
FunctionTestALinkWriteOverTheTimeBudgetStoresNothing
(t *testing.T)
internal/store/commitment_link_write_test.go:164
FunctionTestALinkedHardDeadlineScheduleAnswersLinkedBeforeRenegotiate
---- review findings on the linked-schedule guards ----
internal/store/commitment_links_revalidate_test.go:104
FunctionTestAListStillShowsALinkWithASoftDeletedEnd
(t *testing.T)
internal/store/commitment_links_deleted_end_test.go:100
FunctionTestAMatchingCertificateIsAccepted
TestAMatchingCertificateIsAccepted is the control. Without it, "wrong names are refused" is satisfied by a dialer that refuses everything — which woul
internal/redisdial/redisdial_test.go:121
FunctionTestAMixedPhaseDeploymentDeliversBothWays
(t *testing.T)
internal/events/redis_idspace_test.go:1011
FunctionTestAMoveAcrossWorkspacesDropsTheLinksAndReportsTheCount
(t *testing.T)
internal/store/schedule_carry_links_test.go:47
FunctionTestAMoveOfAnEndedLinkedScheduleStillDropsItsLinks
A schedule that already ended is not ended again by the move, but its links still go: the item leaves the workspace.
internal/store/schedule_carry_links_test.go:88
FunctionTestAMoveToAnotherCollectionInTheWorkspaceKeepsTheLinks
(t *testing.T)
internal/store/schedule_carry_links_test.go:102
FunctionTestAMoveToAnotherCollectionKeepsTheGates
A same-workspace move keeps the gates: it is not a carry.
internal/store/schedule_carry_gates_test.go:90
FunctionTestAMovedBlockerLeavesTheChildBlockedForADecision
Moving a blocker to another workspace archives it here: its gate reads cancelled, the child stays blocked for a decision, and a reader who cannot read
internal/store/commitment_gates_display_test.go:104
FunctionTestANoLongerNeededOccurrenceRecordsNoMissAndALapsingOneDoesNotLapse
(t *testing.T)
internal/store/commitment_cancel_test.go:118
FunctionTestAParentMoveLeavesANoLongerNeededChildWhereItIs
(t *testing.T)
internal/store/commitment_cancel_test.go:338
FunctionTestAPartialRolesPatchOnALinkedScheduleIsGuarded
The partial door reaches the same guard, and a role no edge uses on a schedule that has an edge on another role stays editable.
internal/store/commitment_links_guard_test.go:100
FunctionTestAPassedOverBlockerOccurrenceLeavesTheChildBlockedForADecision
A passed-over occurrence can no longer succeed.
internal/store/commitment_gates_outcome_test.go:234
FunctionTestAPayloadThatDecodesButIsNotOurEventEndsCoverage
(t *testing.T)
internal/events/redis_idspace_test.go:1148
FunctionTestAPayloadWearingThePrefixEndsCoverageThroughTheRealPath
TestAPayloadWearingThePrefixEndsCoverageThroughTheRealPath is the wiring half of the test above (CONVE-19), and it exists because that one only calls
internal/watchevents/heartbeat_test.go:149
FunctionTestAPayloadWearingThePrefixStillEndsCoverage
TestAPayloadWearingThePrefixStillEndsCoverage is the other half of the frame contract. The prefix buys forward compatibility, but it must not become
internal/events/redis_heartbeat_test.go:500
FunctionTestAPayloadWearingThePrefixStillEndsCoverage
TestAPayloadWearingThePrefixStillEndsCoverage is the other half of the frame contract. The prefix buys forward compatibility, but it must not become a
internal/watchevents/heartbeat_test.go:123
FunctionTestAPersistentlyRegressedGenerationIsAcceptedAsANewSpace
(t *testing.T)
internal/events/redis_idspace_test.go:804
FunctionTestAPhaseOneInterleaveDoesNotDropEveryBuffer
(t *testing.T)
internal/events/redis_idspace_test.go:402
FunctionTestAPhaseOneRestartClearsAStaleEpoch
(t *testing.T)
internal/events/redis_idspace_test.go:482
FunctionTestAPlainBlocksLinkIsUnchangedByGates
A plain blocks link means what it always meant: a commitment gate neither reads nor writes item links, and an item link is no gate.
internal/store/commitment_gates_outcome_test.go:502
FunctionTestAPostRegistrationStragglerBelowTheCursorDoesNotEatAnEarlierEvent
TestAPostRegistrationStragglerBelowTheCursorDoesNotEatAnEarlierEvent guards against the same two-spaces mistake the ID-valued ceiling made, committed
internal/events/redis_subscribe_confirm_test.go:969
FunctionTestAPrefixedGarbagePayloadEndsCoverageEndToEnd
TestAPrefixedGarbagePayloadEndsCoverageEndToEnd is the half the classifier test cannot reach: asserting that decodePayload returns an error says nothi
internal/events/redis_heartbeat_test.go:525
FunctionTestAPrefixedPayloadReachesReconciliationThroughTheRealReceivePath
(t *testing.T)
internal/events/redis_idspace_test.go:659
FunctionTestAProbeDoesNotCreditTheSubscriptionThatReplacedIt
TestAProbeDoesNotCreditTheSubscriptionThatReplacedIt is the wiring for the other half of the fence. publishHeartbeats captures the generation, publish
internal/watchevents/heartbeat_test.go:645
FunctionTestAProbeDoesNotCreditTheSubscriptionThatReplacedItsTarget
TestAProbeDoesNotCreditTheSubscriptionThatReplacedItsTarget closes a residual: the generation binding on lastProbeOK had no deterministic test, only t
internal/events/redis_heartbeat_test.go:1077
FunctionTestAQuietInstanceIsNotCycledOnPhase1
TestAQuietInstanceIsNotCycledOnPhase1 is the regression test for the defect BUG-2738's first draft shipped: with no heartbeat and no traffic, a perfec
internal/watchevents/heartbeat_test.go:278
FunctionTestAQuietWorkspaceIsNotCycledOnPhase1
TestAQuietWorkspaceIsNotCycledOnPhase1 is the regression test for a defect in this unit's first draft, and it is the one most worth keeping. Idle det
internal/events/redis_heartbeat_test.go:1718
FunctionTestAReSaveOfAnEndedLinkedSchedulesTimingIsNotATimingEdit
(t *testing.T)
internal/store/commitment_links_revalidate_test.go:155
FunctionTestARealBusMovesTheRealCounter
The bus-to-adapter-to-registry chain end to end: a real MemoryBus, a real EventsObserver, a real registry, so a bus that never calls its observer or a
internal/metrics/events_observer_test.go:176
FunctionTestARearmedReminderIsNotFiredByAnInFlightPass
TestARearmedReminderIsNotFiredByAnInFlightPass A re-arm can move a reminder into the future between the candidate scan and the fire. It clears fired_
internal/store/reminders_test.go:772
FunctionTestAReconnectOnAnIdleWorkspaceIsNotAReset
The guard that keeps pad_event_sequence_resets_total meaningful: a reconnect on a workspace this instance has buffered NOTHING for is not a coverage b
internal/events/redis_reconnect_test.go:168
FunctionTestARecordedMissUnderLapsesIsLapsedEvenWhenTheOccurrenceStayedOpen
A miss recorded under lapses with the occurrence not closed (the engine kept the miss and left it open) reads as lapsed too.
internal/store/commitment_gates_outcome_test.go:209
FunctionTestARecurringBlockerAdvancingNeitherReblocksNorReleases
The gate is held against the occurrence outstanding when it was written. A recurring blocker moving on to its next occurrence can neither re-block wor
internal/store/commitment_gates_outcome_test.go:106
FunctionTestARecurringGatedChildStaysOpenForItsLaterOccurrences
A recurring gated child is gated by schedule: once open it stays open for its later occurrences.
internal/store/commitment_gates_outcome_test.go:146
FunctionTestARedisBusWithASubscriptionAdmits
TestARedisBusWithASubscriptionAdmits is the control: the refusal keys on the empty slot, not on anything the fixture does. Without it the test above w
internal/watchevents/subscribe_refused_test.go:65
FunctionTestARedisBusWithNoSubscriptionRefusesSubscribers
(t *testing.T)
internal/watchevents/subscribe_refused_test.go:43
FunctionTestARejectedSubscribeLeavesTheSlotEmptyAtConstruction
TestARejectedSubscribeLeavesTheSlotEmptyAtConstruction pins BUG-2799's watchevents half: a SUBSCRIBE Redis answers with an ERROR REPLY. The write succ
internal/watchevents/redis_subscribe_rejected_test.go:26
FunctionTestARejectedSubscribeRefusesItsCallerInsteadOfAdmittingIt
(t *testing.T)
internal/events/redis_subscribe_rejected_test.go:95
FunctionTestAReminderWhoseWorkspaceDisagreesWithItsItemIsInert
TestAReminderWhoseWorkspaceDisagreesWithItsItemIsInert No door writes such a row (CreateReminder derives the pair from the item; import maps within t
internal/store/reminders_test.go:1489
FunctionTestARepeatedIDIsTreatedAsBackwards
(t *testing.T)
internal/events/redis_idspace_test.go:1094
FunctionTestAReplacedBufferVoidsTheCallersCoverageClaim
TestAReplacedBufferVoidsTheCallersCoverageClaim covers this case. An ID-space reset during the wait replaces the buffer wholesale. A position in the
internal/events/redis_subscribe_confirm_test.go:1060
FunctionTestAReplacedLoopExitsQuietly
TestAReplacedLoopExitsQuietly pins the distinction the receive loop's own close-arm exists to make. That arm logs an ERROR and moves pad_watchevents_r
internal/watchevents/heartbeat_test.go:217
FunctionTestAReplacementCanItselfBeCycled
TestAReplacementCanItselfBeCycled is the install-stamp rule stated as the consequence that actually matters. A replacement that inherits the OLD stamp
internal/watchevents/heartbeat_test.go:469
FunctionTestARestoreKeepsTheLinksAndMakesThemWritableAgain
(t *testing.T)
internal/store/commitment_links_deleted_end_test.go:111
FunctionTestAResubscriptionEndsThisInstancesCoverage
A Redis failover drops whatever was published while the subscription was down. Before this, PubSub.Channel resubscribed silently and the replay buffer
internal/events/redis_reconnect_test.go:78
FunctionTestAResubscriptionEndsThisInstancesCoverage
The refusal half: a cursor from BEFORE the outage is no longer served, because this instance cannot account for what it missed while it was disconnect
internal/watchevents/redis_reconnect_test.go:187
FunctionTestAResubscriptionOnAnIdleBusStillEndsCoverage
A resubscription with NOTHING in the buffer yet. Every other reconnect test publishes first, so all of them enter dropCoverage with real bookkeeping
internal/watchevents/redis_reconnect_test.go:528
FunctionTestAResubscriptionSignalsEverySubscriberHoldingTheStreamOpen
THE PROPERTY THE UNIT EXISTS FOR: the client holding the stream OPEN across the flap is told, not merely the one that reconnects. A CLI sitting on api
internal/watchevents/redis_reconnect_test.go:278
FunctionTestAResumeCancelledBeforeItStartsIsDeclinedOutright
TestAResumeCancelledBeforeItStartsIsDeclinedOutright is the entry-path twin. A client can be gone before the handler ever calls in — the request conte
internal/watchevents/redis_bus_resume_test.go:381
FunctionTestAResumingJoinerIsToldSyncRequiredAcrossACycle
TestAResumingJoinerIsToldSyncRequiredAcrossACycle covers the case that discriminates. The concern was that a subscriber arriving DURING a cycle gets
internal/events/redis_heartbeat_test.go:300
FunctionTestARoleAnEdgeUsesCannotBeChangedButOthersCan
(t *testing.T)
internal/store/commitment_links_guard_test.go:62
FunctionTestARowWrittenAtTheCapIsStillClaimable
(t *testing.T)
internal/store/event_outbox_bounds_test.go:451
FunctionTestAScriptThatDiedBeforePublishingDoesNotBlockItsRetry
(t *testing.T)
internal/events/redis_idspace_test.go:960
FunctionTestASecondDragUsesTheOverriddenOffsetsMode
---- 5: a second drag takes mode, zone and all-day from the effective offset.
internal/store/commitment_links_review_test.go:79
FunctionTestASeriesRenegotiationPreviewOfALinkedDeadlineIsRefusedToo
(t *testing.T)
internal/store/commitment_links_revalidate_test.go:124
FunctionTestASkippedBlockerOccurrenceLeavesTheChildBlockedForADecision
(t *testing.T)
internal/store/commitment_gates_outcome_test.go:221
FunctionTestASoftDeletedBlockerLeavesTheChildBlockedForADecision
A blocker removed from the workspace cannot succeed: the child stays blocked for a decision, and a restore brings the blocker back as it was.
internal/store/commitment_gates_outcome_test.go:387
FunctionTestASoftDeletedGatedItemKeepsItsGateAcrossARestore
A gated item put in the trash keeps its gate, and so does a restore.
internal/store/commitment_gates_noschedule_test.go:179
FunctionTestAStaleGoroutineCannotDropTheReplacementBuffer
A receive loop can notice its connection died long after the workspace was unsubscribed and resubscribed under it. Without a generation check in the c
internal/events/redis_reconnect_test.go:294
FunctionTestAStaleRevisionIsRefusedBeforeAuthorityAndReason
A stale revision is the first refusal, whatever else the apply would hit.
internal/store/commitment_links_round2_test.go:92
FunctionTestAStragglerAfterUnsubscribeDoesNotRebuildTheBuffer
The straggler: sub.cancel() signals the receive goroutine, it does not join it, so a message already in flight can land after the workspace was droppe
internal/events/redis_resume_coverage_test.go:55
FunctionTestAStragglerCannotEnterTheReplacementsBuffer
TestAStragglerCannotEnterTheReplacementsBuffer is the fan-out half of the generation fence. The straggler test above covers the liveness stamp; nothin
internal/watchevents/heartbeat_test.go:687
FunctionTestAStragglerCannotRefreshTheLivenessOfItsSuccessor
TestAStragglerCannotRefreshTheLivenessOfItsSuccessor pins the generation check on the liveness stamp. stopRedisSubscription only SIGNALS a receive lo
internal/events/redis_heartbeat_test.go:856
FunctionTestAStragglerDoesNotEraseTheCounterBackwardsFloor
(t *testing.T)
internal/events/redis_idspace_test.go:1299
FunctionTestAStragglerFromAnAbandonedIDSpaceIsDiscarded
(t *testing.T)
internal/events/redis_idspace_test.go:542
FunctionTestAStragglerFromAnEndedSubscriptionCannotVouchForTheNewOne
The straggler test above never resubscribes before the stray message lands, so it passes with a fan-out guard that only checks "does SOME subscription
internal/events/redis_resume_coverage_test.go:120
FunctionTestAStragglerFromBeforeAResetIsBoundedByTheNextNewSpaceEvent
(t *testing.T)
internal/events/redis_idspace_test.go:1221
FunctionTestASubscriberArrivingMidEstablishmentWaitsForTheAcknowledgement
TestASubscriberArrivingMidEstablishmentWaitsForTheAcknowledgement covers the overlap between pendingSubs and wsSubs. establishSubscription installs w
internal/events/redis_subscribe_confirm_test.go:644
FunctionTestASuccessfulProbeStillAllowsCycling
TestASuccessfulProbeStillAllowsCycling is the counterfactual: the premise check must suspend detection when we cannot probe, not disable it. Without t
internal/events/redis_heartbeat_test.go:1233
FunctionTestATimerThatLosesToARejectionAdmitsNobody
TestATimerThatLosesToARejectionAdmitsNobody pins the one ordering the integration tests cannot schedule: the rejection and confirmTimeout become ready
internal/events/redis_subscribe_rejected_test.go:206
FunctionTestATimingEditLeavesAGateFixedOnAnotherOccurrenceAlone
(t *testing.T)
internal/store/commitment_gates_rekey_test.go:93
FunctionTestATimingEditLeavesAnOpenGateAlone
(t *testing.T)
internal/store/commitment_gates_rekey_test.go:54
FunctionTestATimingEditLeavesAnOpenGateOnTheEditedOccurrenceAlone
An open gate whose key IS the outstanding occurrence (it opened some other way) is not followed to the new one either.
internal/store/commitment_gates_rekey_test.go:74
FunctionTestATimingEditOfALinkedScheduleIsRefusedAndWritesNothing
(t *testing.T)
internal/store/commitment_links_guard_test.go:25
FunctionTestATimingEditOfTheBlockerRacingAGateWriteLeavesTheGateOnTheCurrentKey
A timing edit of a blocker racing a gate write onto it: whichever goes first, the gate ends on the blocker's CURRENT outstanding occurrence. The queue
internal/store/commitment_gates_pg_test.go:199
FunctionTestATimingEditOfTheBlockerRekeysAPendingGateFixedOnTheEditedOccurrence
(t *testing.T)
internal/store/commitment_gates_rekey_test.go:33
FunctionTestATimingEditThatEndsTheBlockersScheduleFallsBackToTheItemStatus
An edit that leaves the schedule with no outstanding occurrence leaves the gate with no occurrence to wait for: the blocker item's status decides, as
internal/store/commitment_gates_rekey_test.go:140
FunctionTestATimingEditThatKeepsTheKeyLeavesTheGateAlone
An edit that leaves the outstanding occurrence's key as it was changes nothing.
internal/store/commitment_gates_rekey_test.go:120
← previousnext →16,001–16,100 of 28,840, ranked by callers