ZaStoGram_desktop/Telegram/SourceFiles/test
Repository files (latest commit first)
Filename Latest commit message Latest commit date
loop-uh 8d6c0cf950
All checks were successful
Desktop source guards / guards (push) Successful in 6s
Merge upstream Telegram Desktop v7.4.3
Слияние сделал модуль upstream-sync: 346f10fde5 (v7.4.3).
Конфликты разбирал ИИ: да.
2026-11-12 21:30:50 +00:00
..
docs Add a build hook for the localization capture 2026-11-06 09:24:58 +04:00
README.md [ai] Document macOS notification evidence for the harness 2026-10-29 16:19:46 +04:00
test_activation.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_activation.h [ai] Cite code by symbol, not line number, in the harness 2026-10-08 22:24:50 +04:00
test_agent.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_agent.h [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_animation_clock.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_animation_clock.h [ai] Share the harness console-lock reading with macOS 2026-10-09 23:29:29 +04:00
test_box_button.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_box_button.h [ai] Read button and toast states, toast paint and glyph ink 2026-10-26 23:47:47 +04:00
test_capture.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_capture.h [ai] Qualify full-owner captures and bind mapped rectangles 2026-10-26 23:47:47 +04:00
test_chat_menu.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_chat_menu.h Switch everyone to the new chat view by default. 2026-10-28 18:25:24 +04:00
test_chat_show.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_chat_show.h [ai] Place several chat-list messages in or out of view 2026-11-05 17:12:55 +04:00
test_chat_show_place.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_chat_show_place.h [ai] Record the chat-list place helper's executed checks 2026-11-05 17:12:55 +04:00
test_chat_show_self_test.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_clipboard.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_clipboard.h [ai] Name a locked Windows console in harness self-tests 2026-10-08 22:24:50 +04:00
test_console_lock.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_console_lock.h [ai] Share the harness console-lock reading with macOS 2026-10-09 23:29:29 +04:00
test_console_shield.cpp Fix GCC 15 -Wrange-loop-construct in tests 2026-11-12 17:06:37 +04:00
test_console_shield.h [ai] Add console input shield and QPA mouse harness helpers 2026-10-28 17:46:04 +04:00
test_controller_ready.cpp Fix GCC 15 -Wrange-loop-construct in tests 2026-11-12 17:06:37 +04:00
test_controller_ready.h [ai] Resolve test controllers by account and session 2026-10-26 23:47:48 +04:00
test_corner_patch.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_corner_patch.h [ai] Add corner and animation helpers and media traps 2026-10-08 22:24:50 +04:00
test_custom_emoji.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_custom_emoji.h [ai] Give the test harness an always-ready emoji stub 2026-08-21 15:34:29 +04:00
test_expiry_window.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_expiry_window.h [ai] Add typed expiry observations and cutoff windows 2026-10-26 23:47:48 +04:00
test_fixture_fence.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_fixture_fence.h Новые файлы оригинала: include-пути mtproto по раскладке форка 2026-11-03 17:03:07 -05:00
test_fixture_transaction.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_fixture_transaction.h Новые файлы оригинала: include-пути mtproto по раскладке форка 2026-11-03 17:03:07 -05:00
test_full_owner_capture.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_full_owner_capture.h [ai] Ignore the macOS zoom bit of fixed-size test panels 2026-10-26 23:47:47 +04:00
test_gated_stage.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_gated_stage.h [ai] Gate composed stages and add a fixture transaction 2026-10-26 23:47:47 +04:00
test_gram_reconcile.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_gram_reconcile.h [ai] Add wallet fixture reconciliation and secrecy scan helpers 2026-10-08 22:23:59 +04:00
test_history_fixtures.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_history_fixtures.h [ai] Add a regular service-message fixture to the harness 2026-08-18 11:40:55 +04:00
test_hover.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_hover.h [ai] Clear the hover a synthesized click left behind 2026-08-28 13:11:19 +04:00
test_ink.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_ink.h [ai] Read button and toast states, toast paint and glyph ink 2026-10-26 23:47:47 +04:00
test_l10n_seams.cpp Add test seams to hold requests and override app config values 2026-11-06 09:24:58 +04:00
test_l10n_seams.h Add test seams to hold requests and override app config values 2026-11-06 09:24:58 +04:00
test_lang_ids_migration.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_lang_ids_migration.h Added stable language id storage and migration in settings. 2026-10-27 16:12:57 +04:00
test_lang_pack.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_lang_pack.h [ai] Cite code by symbol, not line number, in the harness 2026-10-08 22:24:50 +04:00
test_launch_fuse.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_launch_fuse.h [ai] Report credential-bearing blocked launches only by label 2026-10-26 23:47:47 +04:00
test_layer_root.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_layer_root.h [ai] Cite code by symbol, not line number, in the harness 2026-10-08 22:24:50 +04:00
test_log.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_log.h [ai] Print helper numbers in the telemetry format on request 2026-10-09 20:43:49 +04:00
test_log_lines.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_log_lines.h [ai] Cite code by symbol, not line number, in the harness 2026-10-08 22:24:50 +04:00
test_log_scan_internal.h [ai] Separate publication identity and secret audits 2026-10-20 17:07:54 +04:00
test_mapped_target.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_mapped_target.h [ai] Qualify full-owner captures and bind mapped rectangles 2026-10-26 23:47:47 +04:00
test_marking_read.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_marking_read.h [ai] Share the harness console-lock reading with macOS 2026-10-09 23:29:29 +04:00
test_menu.cpp Fix GCC 15 -Wrange-loop-construct in tests 2026-11-12 17:06:37 +04:00
test_menu.h [ai] Activate accessible custom menu rows by real keys 2026-10-30 19:47:22 +04:00
test_messages.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_messages.h Add sent message watcher to test harness 2026-07-31 19:11:28 +04:00
test_notify_override.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_notify_override.h [ai] Add a local restored notify override to the harness 2026-10-08 22:24:49 +04:00
test_open_handoff.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_open_handoff.h Apply launch warnings to disguised file contents 2026-10-27 14:35:23 +04:00
test_paint_stream.h [ai] Add native gate, paint replay and motion-window evidence 2026-10-26 23:47:47 +04:00
test_palette.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_palette.h [ai] Add chat-list menu attribution and a palette switch 2026-10-15 14:23:06 +04:00
test_panel.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_panel.h [ai] Ignore the macOS zoom bit of fixed-size test panels 2026-10-26 23:47:47 +04:00
test_post_paint.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_post_paint.h [ai] Qualify full-owner captures and bind mapped rectangles 2026-10-26 23:47:47 +04:00
test_press_origin.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_press_origin.h [ai] Add a harness click that ripples at its press point 2026-10-22 22:11:52 +04:00
test_probe.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_probe.h [ai] Print helper numbers in the telemetry format on request 2026-10-09 20:43:49 +04:00
test_publication_identity_scan.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_publication_identity_scan.h [ai] Separate publication identity and secret audits 2026-10-20 17:07:54 +04:00
test_publication_identity_scan_self_test.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_qpa_mouse.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_qpa_mouse.h [ai] Add console input shield and QPA mouse harness helpers 2026-10-28 17:46:04 +04:00
test_qpa_mouse_self_test.cpp Fix GCC 15 -Wrange-loop-construct in tests 2026-11-12 17:06:37 +04:00
test_queued_drag.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_queued_drag.h [ai] Add console input shield and QPA mouse harness helpers 2026-10-28 17:46:04 +04:00
test_rpc_fixture.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_rpc_fixture.h Поправить пути mtproto в тестовой заглушке RPC из апстрима 2026-09-25 15:29:23 +03:00
test_rpc_retry.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_rpc_retry.h [ai] Record MTP retries, space-class reads and N/A rows 2026-09-03 22:26:26 +04:00
test_runner.cpp Let compiled-in scenario modules register with the test runner 2026-11-05 22:00:48 +04:00
test_runner.h Let compiled-in scenario modules register with the test runner 2026-11-05 22:00:48 +04:00
test_scenario.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_secrecy_scan.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_secrecy_scan.h [ai] Separate publication identity and secret audits 2026-10-20 17:07:54 +04:00
test_session_gate.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_session_gate.h [ai] Let scenario stages switch accounts after the session gate 2026-10-30 19:47:22 +04:00
test_style.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_style.h [ai] Settle chat style before harness colour reads 2026-08-14 17:28:43 +04:00
test_text_drop.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_text_drop.h [ai] Add a clipboard-free text drop helper to the harness 2026-10-08 22:24:47 +04:00
test_text_reads.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_text_reads.h [ai] Cite code by symbol, not line number, in the harness 2026-10-08 22:24:50 +04:00
test_toast_capture.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_toast_capture.h [ai] Read button and toast states, toast paint and glyph ink 2026-10-26 23:47:47 +04:00
test_touch.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_touch.h [ai] Add a registered-device touch helper to the harness 2026-11-05 17:12:54 +04:00
test_transfer.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_transfer.h Add document transfer observation test helper 2026-08-10 18:42:49 +04:00
test_via_window.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_via_window.h [ai] Cite code by symbol, not line number, in the harness 2026-10-08 22:24:50 +04:00
test_wallet_intercept.cpp Merge upstream Telegram Desktop v7.4.1 2026-11-07 16:16:36 +00:00
test_wallet_intercept.h Merge upstream v7.3.1 2026-10-15 21:43:47 +03:00
test_wallet_ready.cpp [ai] Correct the wallet refresh helper's stale notes 2026-10-08 22:23:54 +04:00
test_wallet_ready.h Fix key update, add more logging. 2026-10-08 22:23:58 +04:00
test_web_cache.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_web_cache.h [ai] Add a harness helper to clear web document cache 2026-10-08 22:24:47 +04:00
test_wheel.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_wheel.h [ai] Deliver synthetic wheels up the parent ladder 2026-09-24 22:54:00 +04:00
test_widget_attacks.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widget_attacks.h [ai] Tap widget renderer ops and audit the attack corpus 2026-11-05 17:12:55 +04:00
test_widget_bot.cpp [ai] Let widget test messages come via a bot and ask for the wallet 2026-11-05 17:12:56 +04:00
test_widget_bot.h [ai] Let widget test messages come via a bot and ask for the wallet 2026-11-05 17:12:56 +04:00
test_widget_conformance.cpp [ai] Run the paint-policy fixture in the conformance runner 2026-11-05 17:14:32 +04:00
test_widget_conformance.h [ai] Run the paint-policy fixture in the conformance runner 2026-11-05 17:14:32 +04:00
test_widget_conformance_actions.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widget_conformance_actions.h [ai] Run widget conformance sets through the chat list 2026-11-05 17:12:55 +04:00
test_widget_conformance_bridge.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widget_conformance_bridge.h [ai] Run widget conformance sets through the chat list 2026-11-05 17:12:55 +04:00
test_widget_conformance_context.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widget_conformance_context.h [ai] Run v0.2.3 conformance with contexts, bot and pictures 2026-11-05 17:12:55 +04:00
test_widget_conformance_hosts.cpp [ai] Size the conformance runner's window for one column 2026-11-05 17:14:31 +04:00
test_widget_conformance_hosts.h [ai] Size the conformance runner's window for one column 2026-11-05 17:14:31 +04:00
test_widget_conformance_input.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widget_conformance_input.h [ai] Run widget conformance sets through the chat list 2026-11-05 17:12:55 +04:00
test_widget_conformance_paint_policy.cpp [ai] Run the paint-policy fixture in the conformance runner 2026-11-05 17:14:32 +04:00
test_widget_conformance_paint_policy.h [ai] Run the paint-policy fixture in the conformance runner 2026-11-05 17:14:32 +04:00
test_widget_conformance_pictures.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widget_conformance_pictures.h [ai] Wait out the floating date before conformance captures 2026-11-05 17:12:55 +04:00
test_widget_conformance_state.h [ai] Size the conformance runner's window for one column 2026-11-05 17:14:31 +04:00
test_widget_conformance_stop.cpp [ai] Read a widget the desktop stopped as dead in the runners 2026-11-05 17:14:31 +04:00
test_widget_conformance_stop.h [ai] Size the conformance runner's window for one column 2026-11-05 17:14:31 +04:00
test_widget_fixture.cpp Merge upstream Telegram Desktop v7.4.2 2026-11-11 05:06:26 +00:00
test_widget_fixture.h [ai] Size the conformance runner's window for one column 2026-11-05 17:14:31 +04:00
test_widget_host.cpp [ai] Watch stress windows throughout and count replayed frames 2026-11-05 17:14:31 +04:00
test_widget_host.h [ai] Watch stress windows throughout and count replayed frames 2026-11-05 17:14:31 +04:00
test_widget_ops.cpp [ai] Run the paint-policy fixture in the widget harness 2026-11-05 17:14:32 +04:00
test_widget_ops.h [ai] Run the paint-policy fixture in the widget harness 2026-11-05 17:14:32 +04:00
test_widget_paint_policy.cpp Move widgets to the public twidget layout and its grid fixes 2026-11-05 21:00:59 +04:00
test_widget_paint_policy.h [ai] Run the paint-policy fixture in the widget harness 2026-11-05 17:14:32 +04:00
test_widget_stress.cpp [ai] Read a widget the desktop stopped as dead in the runners 2026-11-05 17:14:31 +04:00
test_widget_stress.h [ai] Read a widget the desktop stopped as dead in the runners 2026-11-05 17:14:31 +04:00
test_widget_stress_internal.h [ai] Read a widget the desktop stopped as dead in the runners 2026-11-05 17:14:31 +04:00
test_widget_stress_live.cpp [ai] Size the conformance runner's window for one column 2026-11-05 17:14:31 +04:00
test_widgets.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_widgets.h [ai] Add console input shield and QPA mouse harness helpers 2026-10-28 17:46:04 +04:00
test_window_exposure.cpp [ai] Add a stress kit runner and a Release test harness 2026-11-05 17:12:56 +04:00
test_window_exposure.h [ai] Share the harness console-lock reading with macOS 2026-10-09 23:29:29 +04:00

Telegram Desktop task-test harness

Read this short guide before designing, authoring or recovering a Telegram -testagent overlay. Then read only the references needed by the task and those helpers' headers. The helper index is a lookup tool; neither it nor the detailed references are a required reading bundle.

The permanent harness provides the runner, fixtures, inputs and evidence instruments. The disposable overlay supplies task-specific setup, production entry points and checks. Search the index before recreating a shared facility. Keep the repository's test_scenario.cpp a no-op outside the saved overlay.

The stage contract

  • run acts once on prerequisites already established by an earlier stage.
  • until observes readiness repeatedly; expected product outcomes belong in then or captureAndInspect. Do not turn a mismatch into a timeout.
  • then asserts or acts on the exact object accepted by readiness. Prefer actOnWidget and lifetime guards over resolving another instance.
  • skipReason names an inapplicable prerequisite, never a failing result. timeoutDetails reports the last identity, geometry and state.
  • Compose stages that depend on a fixture prerequisite inside Runner::addGated, so helper-appended stages inherit its first refusal before resolving anything; declare restoration with Runner::addRecovery (gated groups).
  • Assert and log measured values on both passing and failing paths. A helper refusal or undecidable reading is not evidence of product success.

Read the runner contract before writing stages; it includes the narrowly documented activation exception to a pure wait. Startup/session/chats waits are opt-in: select only those the fixture needs.

Every scenario first passes the session gate, which refuses authorized non-Test accounts unless the task explicitly permits production legs. Configure UseTestSession and its restore stage when the fixture needs an account switch. SetupScenario only registers stages, releases and gate options; the gate does not guard actions taken during setup.

Exact objects and actions

Use RTTI-based harness finders, not Qt findChildren<CustomWidget*>(). Visibility alone does not prove identity, ownership or unobscured geometry. Publish exact live objects/actions when descendant searches are ambiguous; keep lifetime guards across stages and preserve a window's concrete type. Read object publication and input contracts. Resolve another account's or a separate window's controller from the window registry with test_controller_ready, never from activation history.

Resolving a chat's message list

Resolve the section that owns the exact message; do not assume every chat uses HistoryWidget or that item->mainView() is the visible element. Read chat-list resolution for message-row work.

Input helpers

Choose the real event receiver and verify its resulting state. Settlement of postponed calls does not wait for timers, network or animations. A synthetic click is not a lasting hover. Read input helpers for input, launch behavior for activation/background painting, and the selected module's reference for clipboard, menu or other special input. Drive grabbing widgets with QPA mouse; keep console input out with the console shield.

Capture and visual assertions

Capture the real paint owner and the complete requested rectangle. Keep readiness separate from the expected pixels; save the accepted frame before asserting on it. Report logical versus image pixels and DPR explicitly. Strict comparisons keep one render owner, origin and extent (strict path); crop a full-owner render only through the checked crop; bind child rects to their coordinate owner; coordinated samplers share a capture scope. Read capture selection, then select only the needed ink, post-paint, clocked-frame or corner reference. Time a whole widget's paint by recording the real owner with paint streams and replaying it in optimized Qt; a Debug paint timing is not a paint cost.

Helper catalog

Find modules by purpose in the helper index. Entries marked Self-test validate the harness; do not add them to every product scenario.

Building one reliable packed scenario

Derive checks from the task and diff; use the most direct real production entry point. Pack compatible checks into one launch; split only for a named incompatible process-lifetime requirement. After a test flaw, repair all visible flaws together. Keep nondestructive states first and closing actions last. Read scenario design and, when useful, the minimal overlay example.

Account fixture secrets

Run only through the workflow's marked test environment. Fixture secrets stay in process: never put them in code, prompts, logs or evidence artifacts. Read account fixtures before using account secrets or a funded wallet, including its required reconciliation and secrecy-scan links. For public-identity or log-coverage checks, read the separate publication audit contract; neither establishes fixture secrecy. A credential the product hands to an OS launcher, such as a token URL, is declared with the launch fuse's private expectation, which matches it exactly and prints only its label and scheme://host. A required missing fixture is a blocker; do not substitute fabricated proof, reset the funded wallet or discard a possibly unrewritten rotation.

Media fixtures and fixture gates

Metadata alone does not establish playable media. A media-dependent scenario must observe real playback and preserve the validated fixture across stages. Read media fixtures before selecting one.

macOS native notifications

Read notification evidence before a run posts macOS native notifications.

Scenario teardown before quit

Register idempotent release of subscriptions, watchers and timers with Runner::onFinish; a final stage can be skipped by timeout or watchdog. Restore asynchronous UI state in a settling stage, not during quit. Read teardown when retaining state or changing the environment. Mutate and restore a server fixture only through the fixture transaction, which reads before writing, waits the server's deadline and keeps a resumable record.

Failure diagnosis

Read the actual logs/captures, classify the failure, and search the diagnostic reference for that signature. It is a lookup history, not a preflight reading assignment. Preserve prior positive evidence; do not weaken an oracle to get past a refusal.

Adding or changing a helper

Follow the documentation format: one short index entry, an on-demand module reference, and a precise source-header contract. Keep this README under 8 KiB; only universal rules and routing links belong here. Examples, platform caveats, self-test procedures and failure histories belong in the linked references. Cite symbols and files, never source line numbers. The full execution workflow remains in the test loop.