MCPcopy Create free account

hub / github.com/PerpetualSoftware/pad / functions

Functions28,840 in github.com/PerpetualSoftware/pad

MethodSubscribeAndReplaySince
(ctx context.Context, sinceID int64)
internal/server/handlers_watch_events_cancel_test.go:29
MethodSubscribeAndReplaySince
(ctx context.Context, workspaceID string, sinceID int64, maxPerWorkspace int)
internal/server/handlers_events_cancel_test.go:48
MethodSubscribeAndReplaySince
(context.Context, int64)
internal/server/handlers_push_publish_result_test.go:53
MethodSubscribeAndReplaySince
(ctx context.Context, workspaceID string, sinceID int64, maxPerWorkspace int)
internal/server/handlers_stream_gap_test.go:50
MethodSubscribeAndReplaySince
(ctx context.Context, sinceID int64)
internal/server/handlers_stream_gap_test.go:66
MethodSubscribeAndReplaySince
SubscribeAndReplaySince — see the Bus interface doc comment for why this exists. Subscribing and reading the replay buffer under the SAME critical sec
internal/watchevents/watchevents.go:665
MethodSubscribeAndReplaySince
SubscribeAndReplaySince atomically registers the subscriber and captures the buffered notifications above sinceID, under the SAME lock the local fan-o
internal/watchevents/redis_bus.go:745
MethodSubscribeIfAllowed
SubscribeIfAllowed atomically checks limits and subscribes.
internal/events/bus.go:762
MethodSubscribeIfAllowed
(context.Context, string, int)
internal/server/handlers_events_failed_test.go:23
MethodSubscribeIfAllowed
(ctx context.Context, workspaceID string, maxPerWorkspace int)
internal/server/handlers_events_cancel_test.go:41
MethodSubscribeIfAllowed
The context is FORWARDED, not replaced with a Background (BUG-2749). This double sits between the handler and the real bus, so swallowing the request'
internal/server/handlers_stream_gap_test.go:45
MethodSubscribeWithOverflow
(itemID string, onOverflow func())
internal/collab/task1273_overflow_test.go:111
MethodSubscriberCount
SubscriberCount returns the number of active subscribers (for testing/debugging).
internal/events/bus.go:1080
MethodSubscriberCount
SubscriberCount returns the number of active local subscribers.
internal/events/redis_bus.go:1668
MethodSubscriptionCycled
()
internal/events/observer_test.go:83
MethodSubscriptionCycled
()
internal/events/observer_test.go:290
MethodSubscriptionCycled
()
internal/events/redis_heartbeat_test.go:72
MethodSubscriptionUnconfirmed
()
internal/events/observer_test.go:53
MethodSubscriptionUnconfirmed
()
internal/events/observer_test.go:288
MethodSubscriptionUnconfirmed
()
internal/events/redis_heartbeat_test.go:71
FunctionSuggestedConventions
SuggestedConventions returns convention titles from the library that match the detected project info. Returns a map of title → customized content.
internal/cli/detect.go:152
MethodSupportsProtocolVersion
(v string)
internal/mcp/chatgpt_security.go:107
MethodSupportsProtocolVersion
(version string)
internal/server/mcp_stream_keepalive.go:206
MethodSwap
(i, j int)
internal/schedule/internal/rrule/rruleset.go:123
MethodSwap
(i, j int)
internal/schedule/internal/rrule/util.go:23
FunctionTerminalStatusPlaceholders
TerminalStatusPlaceholders returns a comma-separated placeholder string and the corresponding args slice for use in SQL IN clauses. Like the *FromSche
internal/models/terminal.go:388
MethodTerminate
(string)
internal/mcp/transport_test.go:23
MethodTerminate
(string)
internal/mcp/bug2772_remote_agent_identity_test.go:37
MethodTerminate
(string)
cmd/pad/cmd_server.go:1365
FunctionTest422SurfacesBodyAndIsNotABudgetError
(t *testing.T)
internal/decision/decision_test.go:362
FunctionTestA
(t *testing.T)
internal/buildtools/testdata/shardpkg/shard_test.go:7
FunctionTestA413IsTooLargeOnTheRemoteTransport
TestA413IsTooLargeOnTheRemoteTransport covers each producer reachable from the catalog (the enumeration is on BUG-2829's trail), by its real body shap
internal/mcp/too_large_test.go:43
FunctionTestA413IsTooLargeOnTheStdioTransport
TestA413IsTooLargeOnTheStdioTransport drives the marker the CLI root writes through the stdio classifier, with the same guidance the remote door gives
internal/mcp/too_large_test.go:132
FunctionTestABarePayloadAfterAdoptionLeavesTheEpochAlone
(t *testing.T)
internal/events/redis_idspace_test.go:457
FunctionTestABlockerStatusUsesTheDefaultOptionsWhenTheCollectionDeclaresNone
With no declared options the workspace default terminal and abandoned lists decide, as everywhere else.
internal/store/commitment_gates_outcome_test.go:332
FunctionTestABlockerWhoseCommitmentHadEndedIsJudgedByItsStatus
A blocker whose commitment had already ended when the gate was written has no outstanding occurrence to fix: its status decides, as for one with no co
internal/store/commitment_gates_outcome_test.go:350
FunctionTestABlockerWithNoCommitmentSucceedsOnATerminalStatusThatIsNotAbandoned
(t *testing.T)
internal/store/commitment_gates_outcome_test.go:300
FunctionTestABrokenClockDoesNotProduceAnUnpublishableEpoch
THE WIRING, which the table above cannot prove: a clock that would have been fatal still publishes a decodable event with a usable epoch. Zero is the
internal/events/redis_generation_guard_test.go:611
FunctionTestACancelPolicyOnlyUpdateMovesNothingAndAsksForNothing
---- 4: a cancel_policy-only update plans no move.
internal/store/commitment_links_review_test.go:50
FunctionTestACancelledCallerDoesNotWaitOutATLSHandshake
TestACancelledCallerDoesNotWaitOutATLSHandshake is BUG-2754's regression test, and the whole reason this package exists. go-redis's default dialer ha
internal/redisdial/redisdial_test.go:212
FunctionTestACancelledCounterReadIsNotReportedAsRedisTrouble
TestACancelledCounterReadIsNotReportedAsRedisTrouble covers the in-flight twin of the entry-path decline. The entry-side decline catches a caller tha
internal/watchevents/redis_bus_resume_test.go:535
FunctionTestACancelledEstablisherNeverStrandsTheNextSubscriber
TestACancelledEstablisherNeverStrandsTheNextSubscriber is the regression test for the sharpest failure this unit can produce, and the one its filing n
internal/events/redis_subscribe_cancel_test.go:479
FunctionTestACancelledParentOccurrenceMarksOnlyTheLinkedChildUnderTheExplicitPolicy
Only the child paired with the cancelled parent occurrence, and only under the explicit policy: the default leaves a child waiting on its parent.
internal/store/commitment_cancel_test.go:63
FunctionTestACancelledResumeIsADepartureNotARefusal
TestACancelledResumeIsADepartureNotARefusal pins the one path that returns a closed channel WITHOUT an error: a caller whose context has already ended
internal/watchevents/subscribe_refused_test.go:112
FunctionTestACancelledResumeStopsPayingForTheSettleWindow
TestACancelledResumeStopsPayingForTheSettleWindow is BUG-2751's regression test. The settle wait used to be bounded by the BUS's lifetime context, ne
internal/watchevents/redis_bus_resume_test.go:317
FunctionTestACascadeIgnoresALinkRemovedWhileItWaited
A cascade works from the links under the locks: a link another transaction removed moves nothing.
internal/store/commitment_links_revalidate_test.go:86
FunctionTestACascadeLeavesASoftDeletedChildUntouchedAndItsLinkCanBeDeleted
A soft-deleted child takes no part in a cascade: its schedule, reminder, overrides and journal stay as they were, and the delete of its link works.
internal/store/commitment_links_deleted_end_test.go:173
FunctionTestAChildFinishingOrReopeningChangesTheRevision
A child that finishes (or reopens) between preview and apply invalidates the preview.
internal/store/commitment_links_review_test.go:407
FunctionTestAClosedBusRefusesSubscribersWithErrBusClosed
(t *testing.T)
internal/watchevents/subscribe_refused_test.go:84
FunctionTestAColdBufferCannotVouchForAResume
BUG-2731's defect, found in this package by a cross-artifact pass on that unit's branch: an empty buffer answering a resume with a non-nil empty slice
internal/watchevents/cold_resume_test.go:23
FunctionTestAColdJoinAcceptsTheAdjacentCursor
(t *testing.T)
internal/events/redis_idspace_test.go:1113
FunctionTestACollidingRepairIsCaughtBySequenceRatherThanEpoch
A REPAIRED GENERATION CAN COLLIDE, AND WHAT CATCHES IT IS THE SEQUENCE, NOT THE EPOCH (BUG-2740). The repair seeds from wall-clock seconds, which is
internal/events/redis_generation_guard_test.go:464
FunctionTestAConfirmedSubscriptionIsNeverLeftMarkedUnconfirmed
TestAConfirmedSubscriptionIsNeverLeftMarkedUnconfirmed covers the interleave where establishSubscription's select chose the timer but the acknowledgem
internal/events/redis_subscribe_confirm_test.go:719
FunctionTestACopyAcrossWorkspacesGetsNoLinksAndTheSourceKeepsHers
(t *testing.T)
internal/store/schedule_carry_links_test.go:70
FunctionTestACorruptedEpochKeyIsRotatedRatherThanEmitted
(t *testing.T)
internal/events/redis_idspace_test.go:852
FunctionTestACorruptedGenerationCounterIsRepairedRatherThanFatal
(t *testing.T)
internal/events/redis_generation_guard_test.go:46
FunctionTestACorruptedSequenceCounterHealsOnThePhaseOnePath
Phase 1 (the assign script, then a plain PUBLISH) runs the same guard: the publish survives and the repair is reported. What it does NOT get is sync_
internal/events/bug2744_seq_guard_test.go:168
FunctionTestACorruptedSequenceCounterHealsVisiblyIntoANewSpace
Phase 2, every shape: the publish SUCCEEDS, starts a new space at id 1, is reported once with its shape, and a cursor from the old space is refused (s
internal/events/bug2744_seq_guard_test.go:89
FunctionTestACounterGoingBackwardsRaisesTheFloor
(t *testing.T)
internal/events/redis_idspace_test.go:369
FunctionTestACounterResetDuringTheReconnectWindowIsStillReported
A COUNTER RESET THAT HAPPENS DURING THE RECONNECT WINDOW. The two conditions were covered separately and their COMPOSITION was not, which is where th
internal/watchevents/redis_reconnect_test.go:564
FunctionTestACreateWhoseEndMovedAwayWhileItWaitedWritesNoLink
(t *testing.T)
internal/store/commitment_links_revalidate_test.go:47
FunctionTestACrossWorkspaceCopyOfAGatedItemGetsNoGatesAndTheSourceKeepsHers
(t *testing.T)
internal/store/schedule_carry_gates_test.go:67
FunctionTestACrossWorkspaceMoveOfAGatedItemDropsItsGatesAndCountsTheReadableOnes
(t *testing.T)
internal/store/schedule_carry_gates_test.go:38
FunctionTestACursorAtTheDiscardedPeakIsRefusedAfterTheClimb
A CURSOR EQUAL TO THE DISCARDED PEAK IS REFUSED ONCE THE NEW SEQUENCE HAS CLIMBED PAST IT (BUG-3206). A counter-backwards reset raises the floor to t
internal/events/redis_discarded_peak_test.go:23
FunctionTestACursorFromAPreviousIncarnationIsAGapEvenWithAWarmBuffer
The incarnation guard, which is the OTHER way a resume can be unservable on this bus (BUG-2736). Separated from the cold-buffer test above because the
internal/watchevents/cold_resume_test.go:58
FunctionTestACursorInsideAnAbandonedIDSpaceIsRefused
A CURSOR INSIDE AN ABANDONED ID SPACE IS REFUSED UNTIL THE NEW SPACE HAS CLIMBED OUT OF IT (BUG-2743, which BUG-2728 is one member of). When the watc
internal/watchevents/redis_bus_ambiguity_test.go:35
FunctionTestADTSGate
TestADTSGate covers the one place a filename participates, in both directions. The gate is on the stdlib having no SPECIFIC format opinion — applicati
internal/attachments/mime_magic_test.go:207
FunctionTestADeleteOfALinkThatVanishedWhileItWaitedIsNotFound
(t *testing.T)
internal/store/commitment_links_revalidate_test.go:73
FunctionTestADeletedBlockerCommitmentNeverReleasesAPendingGate
The gated item's own commitment going takes nothing from the gate, and the blocker's commitment going with a pending gate does not release the child.
internal/store/commitment_gates_noschedule_test.go:148
FunctionTestADisplayReaderWhoCannotSeeTheBlockerStillSeesTheChildHeld
(t *testing.T)
internal/store/commitment_gates_display_test.go:26
FunctionTestADoneBlockerSucceedsEvenAfterARecordedMiss
`done` wins over a recorded miss, and a miss under stays_owed leaves the gate pending: the work is still owed, and a later Done opens it.
internal/store/commitment_gates_outcome_test.go:166
FunctionTestADoneOccurrenceStillWinsOverAnAbandonedBlockerItem
A recorded `done` occurrence is a stored fact with its own actual time, and a gate opens once: abandoning the item afterwards (or before the gate was
internal/store/commitment_gates_outcome_test.go:565
FunctionTestADragBelowAFinishedParentIsKept
A direct drag of a live child records the offset even when its parent is finished.
internal/store/commitment_links_round2_test.go:140
FunctionTestADragBelowAFinishedParentStillStopsTheCascade
The cascade still stops at a finished non-root node, with the drag recorded.
internal/store/commitment_links_round2_test.go:161
FunctionTestAFailedIDAssignmentPublishesNothing
(Arrangement corrected after the split.) Publish lost its local-counter fallback when the ID assignment fails, and the removal is the fix — the fallba
internal/events/redis_uncovered_test.go:35
FunctionTestAFailedProbeAfterASuccessfulOneStillSuspendsDetection
TestAFailedProbeAfterASuccessfulOneStillSuspendsDetection pins the premise check across a probe that succeeds and then starts failing. WHAT THIS DOES
internal/events/redis_heartbeat_test.go:1182
FunctionTestAFailedProbeSuspendsDetection
TestAFailedProbeSuspendsDetection is the mirror image of the failure this feature finds. PUBLISH travels on the client's ordinary pool while the subsc
internal/watchevents/heartbeat_test.go:319
FunctionTestAFailedProbeSuspendsDetectionRatherThanCyclingBlindly
TestAFailedProbeSuspendsDetectionRatherThanCyclingBlindly is the mirror image of the failure this feature exists to find. The detector's inference is
internal/events/redis_heartbeat_test.go:975
FunctionTestAFailedPublishIsNotLoggedByTheBus
BUG-2732: a failed publish is LOGGED BY THE CALLER, rate-bounded (server.publishActivityEvent). The bus used to log every failure itself as well, once
internal/events/redis_unpublished_id_test.go:133
FunctionTestAFailedResubscribeDoesNotReDropCoverageEveryPass
TestAFailedResubscribeDoesNotReDropCoverageEveryPass covers a full-outage behaviour rather than a logic slip. The probe-failure suspension does not c
internal/watchevents/heartbeat_test.go:891
FunctionTestAFailedSubscribeIsReportedAsItselfAndAtOnce
TestAFailedSubscribeIsReportedAsItselfAndAtOnce pins BUG-2764's watchevents half. Before the fix a SUBSCRIBE whose write failed came back as a healthy
internal/watchevents/redis_subscribe_failed_test.go:57
FunctionTestAFailedSubscribeRefusesItsCallerInsteadOfAdmittingIt
TestAFailedSubscribeRefusesItsCallerInsteadOfAdmittingIt is BUG-2764's regression test. Before the fix a SUBSCRIBE whose write failed came back from
internal/events/redis_subscribe_failed_test.go:91
FunctionTestAFailedSubscriptionIsRefusedWithARetryableStatus
TestAFailedSubscriptionIsRefusedWithARetryableStatus is the BINDING assertion for BUG-2764 (CONVE-19): internal/events vouches for the bus returning S
internal/server/handlers_events_failed_test.go:44
FunctionTestAFinishedChildStaysWhereItIsAndStopsTheCascade
(t *testing.T)
internal/store/commitment_links_review_test.go:333
FunctionTestAFinishedRootStillMovesWithItsChildren
Only a non-root finished node leaves the plan: a finished root still moves, and brings its open children with it.
internal/store/commitment_links_review_test.go:451
FunctionTestAFractionalReminderDoesNotFireEarly
TestAFractionalReminderDoesNotFireEarly is the behavioural half — the stored STRING being right is only interesting because of what the tick does with
internal/store/reminders_test.go:730
FunctionTestAFreshConnectionIsNotToldToResync
The control: a genuinely fresh client sends NO header and must NOT be told to resync. Without this leg the fix above would be indistinguishable from r
internal/server/handlers_events_resume_test.go:261
FunctionTestAFreshSubscriberReceivesAnEventPublishedDuringEstablishment
TestAFreshSubscriberReceivesAnEventPublishedDuringEstablishment is the other half of the ceiling's job, and the one an earlier design got wrong. A fr
internal/events/redis_subscribe_confirm_test.go:589
FunctionTestAFreshSubscriptionIsNotCycled
TestAFreshSubscriptionIsNotCycled is the install-stamp rule. A zero lastSeen reads as 1970 and would cycle the replacement on the next pass — forever,
internal/watchevents/heartbeat_test.go:257
FunctionTestAFreshlySubscribedWorkspaceIsNotCycled
TestAFreshlySubscribedWorkspaceIsNotCycled is JOINT RULE 2. Fails without the install-time stamp: a zero lastSeen reads as 1970, so a subscription th
internal/events/redis_heartbeat_test.go:679
FunctionTestAGateCreateWhoseBlockedItemMovedAwayWhileItWaitedWritesNothing
(t *testing.T)
internal/store/commitment_gates_test.go:292
FunctionTestAGateCreateWhoseBlockerMovedAwayWhileItWaitedWritesNothing
---- re-validation under the locks, as for a link write ----
internal/store/commitment_gates_test.go:280
FunctionTestAGateDeleteOfAGateThatVanishedWhileItWaitedIsNotFound
(t *testing.T)
internal/store/commitment_gates_test.go:305
FunctionTestAGateStateCoversEveryGateAndTheDetailOnlyTheReadableOnes
The state covers every gate, the detail only the gates whose blocker the reader can read: a hidden pending blocker holds the child for everyone.
internal/store/commitment_gates_outcome_test.go:437
FunctionTestAGateViewOfAnUnreadableItemIsRefused
(t *testing.T)
internal/store/commitment_gates_outcome_test.go:456
FunctionTestAGateWithAnUnreadableSucceededBlockerShowsNothingAndStillOpens
---- who sees what ----
internal/store/commitment_gates_outcome_test.go:404
FunctionTestAGatedItemThatGetsACommitmentAndLosesItKeepsItsGate
The gated item may place a commitment later, and lose it again: the gate is the item's, so it is the same gate throughout.
internal/store/commitment_gates_noschedule_test.go:63
FunctionTestAGatedItemsCollectionAndStatusLeaveItsGate
Moving the gated item to another collection of its workspace, or changing its status, leaves the gate where it is.
internal/store/commitment_gates_noschedule_test.go:198
FunctionTestAGenuineCounterFailureIsStillReported
TestAGenuineCounterFailureIsStillReported is the control, and the reason the test above is not simply a hole. The first version of this pair was wrong
internal/watchevents/redis_bus_resume_test.go:559
FunctionTestAHealthyGenerationCounterIsIncrementedNotReseeded
The control: a HEALTHY generation counter is incremented, not replaced by the seed. Without this leg the table above passes against a guard that repai
internal/events/redis_generation_guard_test.go:161
← previousnext →15,901–16,000 of 28,840, ranked by callers