Repository navigation
Commit cdb49d9
Consolidated frontend: TM-161/153/151/156/148/8825/8281 (stage batch) (#175)
* feat(TM-161): add emoji reactions to task comments
Adds reactions to comments in TaskComments.vue:
- reaction chips per comment (emoji + count), highlighted when the
current user has reacted; click to toggle
- a "+" picker exposing a default emoji set (DEFAULT_REACTION_EMOJIS)
- optimistic update on toggle with rollback on API failure
New pure, fully-unit-tested helpers in src/utils/commentReactions.ts:
- toggleReaction(reactions, emoji, user) — immutable optimistic toggle
- normalizeReactions(raw, currentUserId) — defensive API normalization
(derives count/reacted from users, drops malformed/zero-count entries)
API layer: toggleCommentReaction(commentId, emoji, currentUserId) in
src/actions/tmgr/comments.ts -> POST /comments/{id}/reactions/toggle,
returns the normalized ReactionSummary[].
Test runner enablement (same as TM-8401, jest was unrunnable on master):
jest.config.js -> .cjs, ts-jest lib:[ES2021] tsconfig, npm test uses
default discovery. 29 tests green; npm run build green.
NOTE: backend (TMBE) must persist reactions and expose the toggle
endpoint + include `reactions` on the comment payload. Frontend degrades
gracefully (empty bar) until the endpoint exists.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-161): ignore stale out-of-order reaction toggle responses (codex P2)
Guard each comment's toggle with a per-comment sequence number so a
slower earlier request can no longer overwrite the state produced by a
later one. Only the latest request applies its server response (or its
rollback).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(TM-153): keyboard navigation of the task list
ArrowUp/ArrowDown move a visible focus between task rows and Enter opens
the focused task (same as clicking it) in TasksListComponent.vue:
- window keydown handler scoped so it does NOT fire while typing in
inputs/textareas/selects/contenteditable/editor surfaces, nor while
any task modal is open
- ignores modified combos (Ctrl/Alt/Meta/Shift) so it never clashes with
App.vue's Ctrl+Alt+Arrow workspace switching
- focus stops sensibly at the ends (no wrap); focused row gets a visible
ring; focus scrolls into view and resets when the list changes
Navigation logic lives in pure, unit-tested helpers
(src/utils/listKeyboardNavigation.ts): nextFocusIndex() and
shouldIgnoreNavigationTarget(). 12 unit tests.
Scope: main tasks LIST view (per task guidance; board left as follow-up).
Test-runner enablement (jest was unrunnable on master): jest.config.js
-> .cjs, ts-jest lib:[ES2021] tsconfig, npm test uses default discovery.
npm test green (32 tests); npm run build green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): don't hijack Enter from focused interactive controls (codex P2)
When a row is focused via the arrow keys, the window-level Enter handler
previously opened the focused task even if real browser focus was on a
button/link/pagination control, breaking keyboard activation of those
controls. Enter now bails when the event target is interactive
(isInteractiveTarget), letting the browser activate the control.
Adds isInteractiveTarget() + 5 unit tests (37 total).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(TM-151): hint to configure category when GitHub repo / Cursor agent unset
When a task's category has no GitHub repository linked or no Cursor agent
configured, clicking the Git activity / Cursor agent buttons now shows an
explanatory dialog with a CTA that navigates to the category settings,
instead of opening a feature that cannot work. When the integration IS
configured (or its status is still unknown), behaviour is unchanged.
- src/utils/categoryIntegrationHint.ts: pure, unit-tested show-logic
(getCategoryIntegrationHint / isCategoryIntegrationConfigured)
- NewForm.vue: loads category GitHub + Cursor status after task load,
openCursorAgent()/openGitActivity() gate the modals, goToCategorySettings()
routes to WorkspaceCategoryEdit; hint Dialog added
- 6 unit tests for the show-logic
Does NOT modify jest.config/package.json (unmerged stack #165/166/167);
tests verified via an out-of-tree jest config. npm run build green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-151): track integration loaded-state per integration (codex P2)
Previously a single categoryIntegrationLoaded flag was set when either
status request fulfilled, so a failed getCursorStatus (or GitHub status)
forced its boolean to false and wrongly showed the hint / blocked the
modal for an integration whose status was actually unknown. Now GitHub
and Cursor each have their own loaded flag, set only when that specific
request fulfils; a hint shows only when that integration's status is
known AND unconfigured.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(TM-148): remove requestDeduplicator, fold single-flight into requestCache
The standalone requestDeduplicator masked the real cause of duplicate GETs:
requestCache only stored *resolved* values, so the cache entry was populated
only after a response returned. Concurrent callers during that in-flight window
all missed the cache and each fired their own request — the deduplicator was a
second, parallel Map patching that gap.
Root-cause fix: make requestCache itself promise-aware (single-flight). New
requestCache.getOrFetch(key, fn, { ttl, cache }) registers the in-flight Promise
synchronously before the first await, so concurrent identical requests coalesce
onto one Promise — there is no miss window left for the deduplicator to cover.
On success it stores the result under ttl; on rejection it drops the in-flight
entry so retries refetch and errors are never cached. `cache: false` coalesces
without persisting.
- Add getOrFetch + clearInFlight to requestCache; clear() now also drops in-flight.
- Migrate all call sites: notifications.getUnreadCount, tasks.getSortedTasksByStatus,
pomodoro.getPomodoroState.
- store/index.js: requestDeduplicator.clear() -> requestCache.clearInFlight()
(logout relies on requestCache.clear() which now clears both).
- Delete src/utils/requestDeduplicator.ts and update the CLAUDE.md data-flow doc.
- Add TDD coverage for coalescing, ttl expiry, null caching, cache:false,
error-not-cached, clearInFlight, and clear.
Note: getPomodoroState previously had a dead `cached !== undefined` early-return
(requestCache.get never returns undefined) that could short-circuit to null without
fetching; routing it through getOrFetch (has()-based presence) restores correct
fetch-on-miss while still caching null results.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-148): prevent poison in-flight entry on synchronous fetchFn throw
Codex review [P2]: if fetchFn threw synchronously, the async IIFE's
`finally { inFlight.delete(key) }` ran during the synchronous phase —
before `inFlight.set(key, request)` — so the rejected promise was then
stored and never cleaned up. Every subsequent call for that key returned
the stale rejected promise and retries never refetched.
Fix: invoke fetchFn eagerly inside try/catch, converting a synchronous
throw into Promise.reject, then build the set/cleanup chain on top. The
`.finally(delete)` now always runs in a later microtask, after the
synchronous `inFlight.set`, so the entry is reliably cleared on both
resolve and reject. fetchFn is still invoked synchronously, preserving
same-tick coalescing.
Adds a regression test: fetchFn throwing synchronously must clear the
in-flight entry and allow the next call to refetch.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(TM-156): active cursor agents indicator with jump-to-task
Adds a header indicator (next to the notification bell, in CustomSidebar)
that lists currently-active cursor/cloud agents (status CREATING/RUNNING)
and navigates to the agent's task on click.
- actions/tmgr/cursor.ts: getActiveCursorAgents() against the expected
global endpoint GET /cursor-agents/active (tolerant of array | {agents} |
{data} shapes, mirroring getCursorAgents).
- utils/cursorAgents.ts: pure, unit-tested helpers — isActiveCursorAgent,
filterActiveCursorAgents, cursorAgentLabel, cursorAgentTaskRoute
(builds the WorkspaceTask route /:workspace_code/tasks/:task_id).
- components/cursor/ActiveCursorAgents.vue: shadcn DropdownMenu indicator
with count badge, RUNNING/CREATING status dots, empty/loading states,
20s polling + reload on open, dark-mode variants. Network errors are
swallowed to an empty list so the UI degrades gracefully until the
backend endpoint ships.
- CustomSidebar.vue: mount the indicator beside NotificationBell.
TDD: src/utils/__tests__/cursorAgents.test.ts (8 tests).
BACKEND DEPENDENCY (FE built on the expected contract, not blocking):
there is no global "active agents" endpoint yet — only per-task
GET /tasks/{id}/cursor-agent. Needs GET /cursor-agents/active returning the
current user's CREATING/RUNNING agents. Reported to PM for a Java ticket.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-156): route jump-to-task via the agent's own workspace
Codex review [P2]: with a global active-agents list, clicking an agent
whose task lives in another workspace navigated to
/:current_workspace/tasks/:task_id, landing on the wrong workspace (or
failing to load the task) in multi-workspace use.
Fix: cursorAgentTaskRoute now prefers the agent's own workspace_code and
only falls back to the current workspace when the agent doesn't carry one.
Add optional workspace_code to the CursorAgent type (populated by the new
GET /cursor-agents/active endpoint). +2 regression tests (cross-workspace
jump; null/empty fallback).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-156): never guess workspace for a global agent jump (codex P2)
Codex round 2: falling back to the current workspace when an agent has no
workspace_code still misroutes cross-workspace agents (the task opens under
the wrong workspace context). For a global list the FE must not assume the
current workspace.
- cursorAgentTaskRoute now returns null unless the agent carries its own
workspace_code (no current-workspace guessing); add canJumpToCursorAgentTask.
- ActiveCursorAgents.vue: agents without a known workspace are shown but
rendered non-jumpable (DropdownMenuItem :disabled, jump arrow hidden);
drop the now-unused current-workspace resolution.
- Tests updated: route built only from the agent's own workspace_code, null
otherwise; canJumpToCursorAgentTask coverage.
This makes workspace_code a REQUIRED field of the GET /cursor-agents/active
contract (reported to PM), not nice-to-have.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor(TM-156): make workspace_code a required field of the active-agents contract (codex P2)
Codex: with the no-guessing route, if the global endpoint returns only the
minimum fields (no workspace_code) every item becomes non-jumpable, defeating
the advertised jump-to-task. Resolve per codex's first suggested option — make
workspace_code part of the REQUIRED backend contract:
- types/cursor.ts: add ActiveCursorAgent (CursorAgent + required workspace_code),
documenting the GET /cursor-agents/active contract.
- actions/tmgr/cursor.ts: getActiveCursorAgents() now returns ActiveCursorAgent[]
with a contract comment.
canJumpToCursorAgentTask / cursorAgentTaskRoute stay as runtime defense-in-depth
against a contract violation, but under the (now required) contract every agent
carries its workspace and the jump always works.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): suppress list nav while modalStack modals are open (codex P2)
isAnyModalOpen only checked currentTaskIdForModal/showCreatingTaskModal/openModals,
so a modal that registers solely in state.modalStack (e.g. TasksMultipleActionsModal)
left the window keydown handler active — ArrowUp/Down/Enter could move focus or open a
task behind the modal. Include modalStack.length in the guard.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-161): per-(comment,emoji) reaction toggles; never clobber concurrent emojis (codex P2)
The seq-guard from 9e8097d was keyed only by comment id, so two quick toggles
of DIFFERENT emojis on the same comment raced: the later response applied and
the earlier successful one was discarded (server had both, UI showed one until
reload).
Fix:
- Key the in-flight sequence guard per (comment, emoji), so toggles of distinct
emojis no longer invalidate one another.
- Replace the wholesale `comment.reactions = server` with
mergeServerReactionForEmoji(current, source, emoji): apply only the toggled
emoji's authoritative state (from the server on success, the pre-toggle
snapshot on error) onto the current list, leaving other emojis untouched.
Adds regression tests for the race (mergeServerReactionForEmoji): a slow emoji
response must not clobber a concurrently-toggled different emoji; targeted
removal/update; immutability.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): track focused task by id so in-place list updates keep focus (codex P2)
focusedIndex is a raw index, but the parent mutates the tasks array in place
(unshift/splice on live updates), so after an insert/removal above the focused
row the ring + Enter targeted the wrong task. Track focusedTaskId and reconcile
the index whenever the task id-sequence changes (focusOrderKey watcher); clear
focus when the task leaves the list.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): scope list keyboard nav to the focused list (codex P2)
The window keydown handler previously fired regardless of where focus was, so
ArrowUp/Down anywhere on the page (body, header, content above the list) moved
the task highlight and swallowed normal key behavior. Make the list root
focusable (tabindex) and only handle keys when focus is within the list
(root.contains(event.target)), so the shortcut no longer leaks outside it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-161): serialize same-emoji reaction toggles per (comment,emoji) (codex P2)
Codex round 2: the per-(comment,emoji) sequence guard fixed cross-emoji
clobbering but left a same-emoji ordering hole — the reaction endpoint is a
STATEFUL toggle, so its final state depends on the order requests are processed
server-side, not the order responses arrive. Rapidly toggling the same emoji
could leave the UI showing the opposite state until reload.
Fix: serialize in-flight toggles per (comment,emoji) via a new tested
keyedSerializer util, so same-emoji requests reach the server in click order and
the last settled response is authoritative; different emojis stay independent
and still merge per-emoji (no cross-emoji clobber).
- src/utils/keyedSerializer.ts: createKeyedSerializer() — per-key async
serialization; rejects don't break the chain; no unhandledRejection for
fire-and-forget callers.
- src/utils/__tests__/keyedSerializer.test.ts: ordering, cross-key overlap,
reject-continues.
- TaskComments.vue: drop the per-comment seq guard; optimistic flip per click,
then serializeReaction(`${comment.id}:${emoji}`, …) running the merge.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-161): single in-flight reaction toggle per (comment,emoji); drop queue (codex P2)
Codex round 3: serializing toggles fixed successful ordering but a FAILED
earlier same-emoji request left queued toggles flipping a server state that
never changed (add-then-remove with the add failing => server ends with the
reaction though the user's last click removed it). Driving a stateful toggle
endpoint with a client queue is inherently unsafe.
Fix: allow only ONE in-flight toggle per (comment,emoji). A second rapid click
on the same emoji is ignored until the first settles, so every request acts on
settled, known server state — no queue to diverge on failure. Different emojis
stay independent (different keys) and still merge per-emoji, so they never block
or clobber each other.
- TaskComments.vue: reactionInFlight Set guard; remove the serializer/queue.
- Remove now-unused keyedSerializer util + tests.
- Keep mergeServerReactionForEmoji (+ the cross-emoji race regression test).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): roving-tabindex so Enter targets the highlighted row (codex P2)
Remaining codex P2 (Enter vs DOM-focus desync): arrows only moved the visual
focusedIndex, so when DOM focus sat on a card/button elsewhere, Enter activated
that control instead of opening the highlighted row.
Design call — roving tabindex: rows get tabindex="-1" and arrow-nav moves real
DOM focus onto the highlighted row (in scrollFocusedIntoView, focus({preventScroll})).
Now DOM focus == visual focus, so Enter on the row opens that row; if the user
Tabs to an in-row control, focus is on the control and Enter activates it
(isInteractiveTarget already handles that). Precedence: an explicitly focused
in-row control wins; otherwise Enter opens the highlighted row.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-161): accept raw-array reaction responses (Java shape) (codex P2)
Codex: toggleCommentReaction destructured `{ data: { data } }`, so when the
endpoint returns a raw array (the Java-style shape other endpoints already
handle) `data` was undefined and normalizeReactions returned [], making the
reaction vanish right after a successful toggle.
Fix: add a tested extractReactionsPayload(body) that accepts a raw array or a
{ data } / { reactions } wrapper, and route the toggle response through it
before normalizing. +5 regression tests for the response shapes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): use explicit listRoot ref instead of $el (codex P1)
TasksListComponent is multi-root (<div> + <Confirm>), so Vue 3 exposes $el as a
fragment placeholder, not the list element. That made root.contains(target)
never match and $el.querySelector() find no rows — so ArrowUp/Down/Enter
navigation (and the roving-tabindex focus) silently did nothing.
Fix: bind ref="listRoot" on the list container and use this.$refs.listRoot for
both containment (handleListKeydown) and row lookup/focus (scrollFocusedIntoView).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-153): start list nav at page level without pre-focusing (codex P2)
Codex: the containment guard (focus must be inside listRoot) meant ArrowUp/Down/
Enter did nothing until the user clicked or tabbed into the list — so on initial
load, after a modal closes, or with focus in surrounding layout, the advertised
page-level task-list navigation never started.
Fix: also handle the keys when there is no meaningful focus (event.target is
<body>/<html>), in addition to focus inside the list. Focus parked inside some
other element is still left alone, and inputs/editors remain excluded via
shouldIgnoreNavigationTarget. TasksListComponent renders once per page (list view
and category view both bind a single instance), so there is no multi-list
conflict from page-level handling.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-148): align task-status cache key with the invalidation pattern (tmgr #8827)
getSortedTasksByStatus cached under requestCache.generateKey(`tasks/status/${id}`)
(slash), but every mutating task action invalidates with /^tasks-status-/
(hyphen). The pattern could never match the key, so board results cached via
getBatchTasksByStatuses (useCache=true, 30s TTL) stayed stale for up to ~30s
after create/update/delete/reorder.
Fix: generate the cache key in the `tasks-status-` namespace the 8 invalidation
sites already use (the HTTP URL stays `tasks/status/...`). +3 regression tests
covering the alignment, the legacy-key non-match, and the namespace.
Follow-up from the TM-148 review (PR #169). jest.config left untouched
(unmerged stack) — tests run via inline --config.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-151): reload integration status when the category changes (codex P2)
The hint used categoryHasRepository/categoryCursorConfigured cached for the
category loaded at open time; changing form.project_category_id via either
category combobox did not refresh it, so after moving a task between a
configured and unconfigured category the hint could be wrong. Watch
form.project_category_id and reload the integration state on change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-151): ignore stale integration-status responses on category change (codex P2)
If the category changes while a loadCategoryIntegrationState() request is in
flight, the older response could resolve later and overwrite the state for the
wrong category. Capture the requested categoryId and bail after the await when
form.project_category_id no longer matches.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(TM-8825): lock frontend jest test runner; drop dead Vue scaffold spec
Tech-debt follow-up to TM-8401 (PR #165). Master's test runner was broken:
jest.config.js used module.exports under "type":"module" (fails to load),
and `npm test` pointed at a nonexistent ./tests path.
- Rename jest.config.js -> jest.config.cjs so the CommonJS config loads
cleanly under the package's ESM type; pin ts-jest target/lib (ES2021).
- testMatch limited to TS specs (**/__tests__/**/*.test.ts, *.spec|test.ts).
- Remove src/components/Index/Index.spec.js: a vue-cli scaffold stub with
only commented-out, zero-assertion cases. Consciously deleted rather than
wiring vue-jest/babel to run a no-op (per TM-8825 scope).
- Fix package.json test script: `jest --verbose ./tests` -> `jest --verbose`.
Validation: npx jest --verbose -> 20/20 passing (dashboard-types.test.ts).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(TM-8281): hours+minutes overtime in lists + status-column tooltip
Finish the task-overtime display:
- Extract a reusable formatOvertime(seconds) util ('1h 23m' / '23m' / '1h', null
when <=0 or sub-minute) so sub-hour overtime stays visible instead of rounding
to '0.0h'. Reuse it in getTaskOvertime (DRY, identical output).
- selectedSummary now shows '+1h 23m overtime' instead of hours-only '+X.Xh'.
- Add a hover tooltip (title) on the per-row time in the status column showing
that row's overtime (e.g. '+1h 23m overtime').
Unit tests for the formatter (sub-minute/sub-hour/whole-hour/combined).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>1 parent 8c5f54c commit cdb49d9
29 files changed
Lines changed: 2757 additions & 858 deletions
File tree
- src
- actions/tmgr
- components
- Index
- cursor
- general
- tasks
- pages
- store
- types
- utils
- __tests__
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
28 | 28 | | |
29 | 29 | | |
30 | 30 | | |
31 | | - | |
32 | | - | |
33 | | - | |
| 31 | + | |
| 32 | + | |
34 | 33 | | |
35 | 34 | | |
36 | 35 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1 | 1 | | |
| 2 | + | |
| 3 | + | |
2 | 4 | | |
3 | 5 | | |
4 | 6 | | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
5 | 10 | | |
6 | 11 | | |
7 | 12 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1 | 1 | | |
2 | 2 | | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
3 | 8 | | |
4 | 9 | | |
5 | 10 | | |
6 | 11 | | |
7 | 12 | | |
8 | | - | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
9 | 17 | | |
10 | | - | |
| 18 | + | |
11 | 19 | | |
12 | 20 | | |
13 | 21 | | |
| |||
60 | 68 | | |
61 | 69 | | |
62 | 70 | | |
| 71 | + | |
| 72 | + | |
| 73 | + | |
| 74 | + | |
| 75 | + | |
| 76 | + | |
| 77 | + | |
| 78 | + | |
| 79 | + | |
| 80 | + | |
| 81 | + | |
| 82 | + | |
| 83 | + | |
| 84 | + | |
| 85 | + | |
| 86 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1 | 1 | | |
2 | | - | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
3 | 8 | | |
4 | | - | |
5 | | - | |
6 | | - | |
7 | | - | |
8 | | - | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
9 | 17 | | |
10 | 18 | | |
11 | | - | |
12 | | - | |
13 | | - | |
14 | | - | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
15 | 28 | | |
16 | 29 | | |
17 | 30 | | |
18 | | - | |
19 | | - | |
20 | | - | |
21 | | - | |
| 31 | + | |
| 32 | + | |
| 33 | + | |
| 34 | + | |
22 | 35 | | |
23 | 36 | | |
24 | | - | |
25 | | - | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
26 | 52 | | |
27 | 53 | | |
28 | 54 | | |
29 | | - | |
30 | | - | |
31 | | - | |
| 55 | + | |
| 56 | + | |
| 57 | + | |
32 | 58 | | |
33 | | - | |
34 | | - | |
35 | | - | |
| 59 | + | |
| 60 | + | |
| 61 | + | |
36 | 62 | | |
37 | 63 | | |
38 | | - | |
39 | | - | |
40 | | - | |
| 64 | + | |
| 65 | + | |
| 66 | + | |
| 67 | + | |
| 68 | + | |
| 69 | + | |
| 70 | + | |
| 71 | + | |
41 | 72 | | |
42 | 73 | | |
43 | | - | |
44 | | - | |
45 | | - | |
46 | | - | |
| 74 | + | |
| 75 | + | |
| 76 | + | |
| 77 | + | |
| 78 | + | |
| 79 | + | |
| 80 | + | |
47 | 81 | | |
48 | 82 | | |
49 | 83 | | |
50 | | - | |
| 84 | + | |
51 | 85 | | |
52 | 86 | | |
53 | | - | |
54 | | - | |
55 | | - | |
| 87 | + | |
| 88 | + | |
| 89 | + | |
| 90 | + | |
| 91 | + | |
56 | 92 | | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1 | 1 | | |
2 | 2 | | |
3 | | - | |
4 | 3 | | |
5 | 4 | | |
6 | 5 | | |
| |||
60 | 59 | | |
61 | 60 | | |
62 | 61 | | |
63 | | - | |
| 62 | + | |
64 | 63 | | |
65 | 64 | | |
66 | 65 | | |
| |||
70 | 69 | | |
71 | 70 | | |
72 | 71 | | |
73 | | - | |
74 | | - | |
75 | | - | |
76 | | - | |
77 | | - | |
78 | | - | |
79 | 72 | | |
80 | | - | |
81 | | - | |
82 | | - | |
83 | | - | |
84 | | - | |
85 | | - | |
86 | | - | |
| 73 | + | |
| 74 | + | |
| 75 | + | |
| 76 | + | |
| 77 | + | |
| 78 | + | |
| 79 | + | |
| 80 | + | |
87 | 81 | | |
88 | 82 | | |
89 | 83 | | |
| |||
102 | 96 | | |
103 | 97 | | |
104 | 98 | | |
105 | | - | |
106 | | - | |
107 | | - | |
108 | | - | |
| 99 | + | |
| 100 | + | |
| 101 | + | |
| 102 | + | |
| 103 | + | |
109 | 104 | | |
110 | 105 | | |
111 | | - | |
| 106 | + | |
112 | 107 | | |
113 | 108 | | |
114 | 109 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1 | 1 | | |
2 | 2 | | |
3 | 3 | | |
4 | | - | |
5 | 4 | | |
6 | 5 | | |
7 | 6 | | |
| |||
49 | 48 | | |
50 | 49 | | |
51 | 50 | | |
52 | | - | |
53 | | - | |
54 | 51 | | |
55 | | - | |
56 | | - | |
57 | | - | |
58 | | - | |
59 | | - | |
60 | | - | |
61 | | - | |
| 52 | + | |
| 53 | + | |
| 54 | + | |
| 55 | + | |
| 56 | + | |
| 57 | + | |
| 58 | + | |
| 59 | + | |
| 60 | + | |
| 61 | + | |
62 | 62 | | |
63 | 63 | | |
64 | 64 | | |
| |||
0 commit comments