Keyboard-launcher launches (Raycast/Spotlight Enter) left the
LSUIElement app unforegrounded with a missing Dock icon until Mission
Control forced a refresh. Investigation across real-machine rounds
(recorded in the ADR-0017 supplements) pinned three cooperating causes:
- The launch activation retry never started at launch: the workspace
window's participation events (anchor attach / occlusion / key) could
all miss the launch moment when content installed before the window
became visible and the visibility notification fired before the
anchor attached. A deferred visibility rescan plus re-armable chain
(until truly settled) close this gap.
- NSApp.isActive can report an activation illusion while the menu bar
still belongs to the previous frontmost app, so the settle check now
requires both isActive and menu bar ownership.
- A same-turn accessory->regular bounce is coalesced away by the
system; forced activation is a three-hop sequence with real settle
delays (accessory -> 0.1s -> regular -> 0.15s -> activate), which
re-registers the app with WindowServer/Dock and reclaims the menu
bar. Each acting attempt also re-asserts makeKeyAndOrderFront and
orderFrontRegardless as a visible-layer fallback.
Diagnostics: launchActivationRetry and windowAnchorAttached events
(unified-log category "window") record per-attempt facts (activation,
local policy, frontmost app, menu bar owner, space, window visibility);
user-environment facts stay out of shared diagnostic reports.
Replace the custom encrypted container with AgeSwift 0.1.0 so standard
age tools can inspect sensitive configuration independently of the app.
Use independent Keychain X25519 identities for catalog.age and
proxy-listening.age. Preserve atomic commits and confirmed recovery;
reject historical NG2 formats without migration or cleanup.
Move InMemoryDiagnosticReportExporter and InMemoryConfigurationGroupFileExporter
from App/PlatformEffects to Tests/Support: both are test doubles with no
production caller, and production composition wires only the AppKit adapters.
Delete RuleRuntimeCompiler.validation, a wrapper whose only callers were two
assertion blocks in the controller deployment test; those restated
custom-to-custom shadowing and validation order-independence, both pinned by
CustomRuleValidatorTests and RuleAnalysisTests. RuleRuntimeCompiler.compile
stays and keeps its production caller.
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
Six sites re-asserted the predicate waitUntilRuntimeSettles already checks,
with no suspension point in between, so they could never fail; their intent
messages are folded into comments where one was not already stated.
Two commit-failure cases asserted the fake's bookkeeping (updateCalls stays
empty) which the fake guarantees by throwing before recording; the outcome
and committed-snapshot assertions already prove the real behavior.
Also drop a no-op loop that mutated a discarded struct copy, a duplicated
isWellFormed line, two WorkspaceRoute cases subsumed by the constructor and
the publish-count case, a duplicated presentation-message assertion, and the
round-trip halves of two payload-shape cases already covered case-by-case.
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
Each removed assertion holds by construction and can never fail:
- jsonData() compared with itself cannot detect cross-run byte stability;
bindAddress and mode are literal-returning computed properties
- RuleAction/RuleSourceKind allCases assertions pin synthesized enum order
(neither allCases list is used in production); the persisted action tokens
stay pinned by the kept rawValue assertions and source-kind tokens by
fixture decoding through RuleSnapshotStore
- memberwise-constructor echo in RuleSourceIdentity, fixture-guaranteed
name == id rows, a share draft field pair built from one string, and a
local struct compared to its own assignment
- the synthesized-Codable round trip through literal JSONEncoder/JSONDecoder
The duplicate-import assertion becomes a real check: Set of node ids instead
of a count that restated roots.count, which is what its comment claims.
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
Every case here restates a rule another test already pins with the same or a
wider input class:
- the activation group case is byte-identical to its neighbour, minus an extra
assertion; the rule snapshot version guard is re-driven with a second value
of the same schemaVersion check; the global-mode deployment-failure case is
a mode-literal copy of the strictly stronger direct-mode one; the clear-
failure case duplicates the deferred-retry sibling
- two RuntimeDocument projection cases only re-map locals, already asserted as
full array equality; the IPv6-LAN endpoint case repeats the direct-mode case
(the derivation reads only ports)
- the three HTTP export cases varied only listenerMode while the derivation
ignores it and one name contradicted its own expectation; one loop over all
listener modes now states the real rule
- the 0700/0600 permission baseline is one AtomicFileWriter behavior and stays
pinned through RuntimeFileStoreTests, which also covers self-heal and atomic
replace
The command-boundary case keeps its mixed-scenario isolation check (four
commands, four independent adapter counters) and drops only its unfalsifiable
queriedTargetIDs claim.
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
testExternalSettingsChangesNeverSupplyAFallbackOrAlterThreeStateResults
mutates a preferences fixture 12 ways but never passes it to RulesWorkflow
(which takes no preferences) and the matcher path is a pure function of the
collection; the 48 assertions are four matcher outcomes repeated 12 times and
each outcome is covered elsewhere.
The two ACL "does not load custom rules" cases build a customRules array that
ProxyACLDocument.global/direct cannot receive (both take only a URL), so their
contains() negatives could never fail; the template shapes are pinned by
ProxyACLTests and the behavioral claim by CustomRuleControllerTests.
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
Four listening-file failure cases inject InMemoryProxyListeningStore.saveError
and assert saved == nil, which the fixture guarantees because save throws
before assigning; the only production save call is updateListening and none of
these paths reaches it. Their behavioral content is covered by sibling tests,
so keep only the three assertions that were unique there (agent intent
persistence, runtime stopped after the off cascade, no commit in flight) and
move them into those siblings.
Same pattern in testProxyExceptionsIgnoreListeningPersistenceFailure: the
path calls only the non-throwing updateProxyExceptions, so its updateError
injection is inert and the rest duplicates the sibling case. The plugin
diagnostic test fed an unproducible fixture (user-sourced facts never carry a
version, since managedInfo exists only for managed sources); assert the
reachable property, the source label in the report row. FakeSettingsSaveError
and the now-unused saveError seam go with them.
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
Three tests claimed coverage they could not deliver:
- subscription URL credential cleanup asserted a lookup derived from the
already-empty subscription list, so it held even on a leak; production has
no separate URL credential store (SubscriptionRecord.url is a plain field)
- the runtime log event sweep planted no secret values, so its secret half
could never fail; keep the live JSON-dump check and rename the test to match
- clipboard and QR saver tests exercised only in-memory doubles that live in
production source and have no other caller; delete the six tests and the
three doubles (AppKit adapters stay untested, as before)
Co-Authored-By: DeepSeek Harness <noreply@deepseek.com>
The root system-proxy helper logged nothing and launchd discards its
stdout/stderr, leaving helper launches, connection refusals, GUI-side
timeouts and root-side failures invisible. Share the minimal unified-log
emission unit (SystemLog) across GUI, wrapper and helper, and add a
dedicated closed helper event enum (lifecycle minimum: listener ready,
connection accepted/rejected with peer pid, typed apply/clear results)
with ADR-0045 privacy splitting. The emission seam is injected at the
helper engine and listener for test doubles. Helper events are
Console-only: no stderr, no agent.log, no diagnostic view or report
surface (ADR-0047).
Collapse the four definitions of "one log line" into LogLine (timestamp,
level, source, text) with a single renderer and parser. agent.log lines
now carry the uniform prefix, absorbing the ssxng: event marker into the
source column; legacy lines degrade to raw display without rewriting.
Level detection moves into the row model with a classic three-level
taxonomy whose unified-log projection stays unchanged.
The diagnostics log view merges GUI events and the agent.log tail through
DiagnosticLogIntake into one time-ordered list with level and source
filters; the tail window logic moves into the intake and blank lines are
kept for raw fidelity. The report whitelist projection and unified-log
routing are unchanged.
Qiu Yuzhou
