Activity redesign: responsive split view, failure containment, review hardening (follow-on to #2388)

[myieye]

This branch is #2388 (the activity view rework — readable summaries + before/after diff, resolves #2170) plus a deep review-and-polish pass over the whole feature, with develop merged in. Merging this supersedes #2388; nothing here changes the feature's direction.

The layout question

@myieye wasn't fully convinced by the list+detail panels, and @hahn-kev showed the layout crushing the list to 8rem at narrow widths. I researched how version-history UIs handle this before touching it: tools where each item is a rich before/after of one artifact (Google Docs version history, Figma, Notion page history, VS Code Timeline) all converge on list + persistent diff pane, while list-only feeds (GitHub commits, Jira) fit events that are individually small or need their own URL. NN/g's accordion guidance argues against inline expansion for complex, comparison-heavy content, and Material 3 / Microsoft / PatternFly all say the same thing about narrow widths: don't shrink both panes, switch to single-pane at a breakpoint.

So the panel stays, and the narrow-width problem is fixed the way this app already fixes it in Browse:

• Desktop: resizable split panes (same ResizablePaneGroup as Browse, 35/65 default) instead of the fixed minmax(8rem,25%) grid. Newest commit auto-selects, so the page opens answering "what changed recently?".
• Below the mobile breakpoint (768px — also covers the narrow-window screenshot from review): full-width list; tapping a row opens the detail full-screen with a close button; detailOpen is a query param like Browse's entryOpen, so the platform back button closes it. Nothing auto-selects on mobile.

Robustness

The "opening this row breaks the whole activity page" report is now contained twice over: the view is wrapped in the app's ViewErrorBoundary, and each change card's {#await} has a {:catch} that renders an inline "Failed to load this change" card instead of leaving a blank 700px row (template catch, not try/catch — the global-handler rule is respected). I could not reproduce the original crash on this branch (deleting an entry and opening its row works — screenshot below), so the root cause was likely fixed during #2388's iteration; the containment is there for whatever the next bad row is. The homograph-patch summary from review now reads "wulu₁ · Set homograph number to 1".

API / backend changes

• HistoryService.ProjectActivity returns Task<ProjectActivity[]> instead of IAsyncEnumerable — the label resolver already materializes the page and every consumer did too (flagged in Rework the activity view: readable summaries and a before/after diff #2388's own notes). Wire shape is unchanged; the JS interop signature changed with regenerated types.
• ChangeTypesCodeGenerator now calls HistoryService.GetChangeTypeKeyFromType (made public) instead of duplicating the TypeName reflection, so the generated ChangeType union can't drift from the runtime $type discriminators.
• The resolver no longer emits a hardcoded English "sense" subject for a label-less sense; it degrades to null like every other unresolved label and the frontend omits the token (new backend test pins this).

Cleanups

Deleted the storybook-only DiffValue component; DiffRichText now delegates to DiffMultiString instead of duplicating its layout; ActivityListViewOptions renamed to ActivityViewPicker (it only picks the view — the Simple/Detailed list mode it was named for no longer exists, and the two stale references to that mode are gone); the audio diff's added/removed boxes carry screen-reader labels (colour was the only cue); changeType() is exported from change-summary instead of re-implemented in ActivityItem.

Test hardening

• vocab-diff-coverage now parses DiffVocabPrimitive's actual present('…') rows instead of a hand-maintained set — dropping a row now fails the test (the direction that would have caught the historical MorphType-description bug), with a reverse stale-row check.
• Table-driven describeChange cases for the payload-picking handlers that had none (ReplacePublicationChange reads newPublication where Add reads publication — exactly the kind of thing that deserved a pin), translations, component links, vocab patch clears.
• Pinned the deliberate recognizer asymmetry: an entry-creation commit collapses even when mixed with same-root edits; a sense-creation commit must be pure.
• New backend test for the null sense-subject degrade.

Test evidence: LcmCrdt.Tests HistoryServiceActivityTests 36/36 pass; viewer vitest for activity + diff-view + coverage suites 90+ tests pass; svelte-check and eslint clean; FwLiteShared builds with regenerated types committed (no drift); i18n extraction round-trips cleanly, and new msgids got translator-context comments per I18N_CONTEXT_GUIDE (30 added, 6 judged self-explanatory).

E2E: ran FwLiteWeb + vite against the example project, made real edits (field edits, new sense, homograph pair, entry deletion) and exercised the activity view in both layouts with zero page errors.

Screenshots

Desktop split view (delete-commit selected, deletion wash, grouped rows with the homograph line):

``

Collapsed create-entry tree card:

``

Mobile: full-width list → tap → full-screen detail with close:

History dialog from a change card:

(Screenshots were committed at cc47568 and reverted in the next commit so they don't live on the branch.)

Notes for reviewers

• The three red argos statuses are the expected "changes detected — waiting for your decision" baseline approvals: this PR deliberately changes the activity view's visuals (4 UI + 7 e2e snapshots). All GitHub Actions workflows are green (the one FW Lite failure was a vstest testhost-connection timeout on a slow Windows runner; the re-run passed).
• The QueryParamStateBool config snapshots IsMobile.value at construction (back-button semantics won't retrack if the viewport crosses the breakpoint mid-session). This faithfully mirrors Browse's entryOpen — a pre-existing pattern now shared by two views; worth a small shared fix someday, not in this PR.
• No Playwright coverage for the activity view: the in-memory demo project has no history service, so the infra-free UI harness can't drive this page. Follow-up idea: give the demo a stub IHistoryServiceJsInvokable.
• Known-good notes carried over from Rework the activity view: readable summaries and a before/after diff #2388 still apply (@sanity/diff-match-patch dependency, sentence-frame i18n tradeoff, WS-code chips app-wide).

🤖 Generated with Claude Code

https://claude.ai/code/session_01Vm9mEUzcfgX4oUPAbhcfsj

[coderabbitai]

Important

Review skipped

Auto incremental reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: c12ed5ac-1f8e-4744-afb4-61d8177317bf

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

📝 Walkthrough

Walkthrough

Changes

Backend activity enrichment

Layer / File(s)	Summary
Materialized activity records and batch metadata backend/FwLite/LcmCrdt/HistoryService.cs, backend/FwLite/LcmCrdt/ActivityChangeInfoResolver.cs	ProjectActivity now returns enriched arrays with resolved subjects, targets, headwords, and previous snapshots.
Generated activity contracts backend/FwLite/FwLiteShared/TypeGen/*, frontend/viewer/src/lib/dotnet-types/generated-types/LcmCrdt/*	Generated TypeScript types include change metadata, previous snapshots, and known change-type literals.
Backend validation backend/FwLite/LcmCrdt.Tests/*	Tests cover activity filtering, materialization, timestamps, snapshot recovery, and change-info naming across entity and relationship changes.

Frontend activity and diff experience

Layer / File(s)	Summary
Structured summaries frontend/viewer/src/lib/activity/change-summary.*, fact-glyph.ts	Backend changes are converted into typed facts, grouped by root entry, categorized, collapsed, and capped for list rows.
Before/after previews frontend/viewer/src/lib/activity/ActivityItemChangePreview.svelte, frontend/viewer/src/lib/entry-editor/diff-view/*	Activity previews render before/after diffs for entries, senses, examples, vocabulary objects, media, selections, lists, and writing-system values.
Activity navigation and rendering frontend/viewer/src/lib/activity/ActivityView.svelte, ActivityItem.svelte, ListDetailLayout.svelte, BrowseView.svelte	Activity and browse views use responsive list-detail layouts, grouped rows, lazy tree previews, close controls, commit details, and mobile URL state.
Supporting activity components frontend/viewer/src/lib/activity/{AuthorLabel,ChangeSummary,EntryHeadwordButton,PreviewViewScope,Collapsed*}.svelte, author-color*, assemble-entry.*	Author coloring, historical entity assembly, collapsed previews, entry actions, and view-scoped rendering are added.

Validation and localization

Layer / File(s)	Summary
Coverage tests and stories frontend/viewer/src/lib/activity/*.test.ts, frontend/viewer/src/lib/entry-editor/diff-view/*.test.ts, frontend/viewer/src/stories/**/*	Tests and stories cover summary handlers, historical assembly, author colors, diff rendering, entity coverage, and representative preview states.
UI and locale support frontend/viewer/src/lib/components/*, frontend/viewer/src/locales/*.po	Writing-system display components, homograph controls, layout details, and activity/diff localization entries are updated across supported catalogs.

Estimated code review effort: 5 (Critical) | ~120 minutes

Possibly related PRs

• sillsdev/languageforge-lexbox#1804: Overlaps the HistoryService.ProjectActivity and activity payload changes.
• sillsdev/languageforge-lexbox#2097: Modifies the activity preview context and rendering pipeline.
• sillsdev/languageforge-lexbox#2037: Touches the shared activity-view data shape and list rendering path.

Suggested reviewers: rmunn, hahn-kev

Poem

I’m a rabbit diffing change by change,
With carrots sorted, neat and strange.
Before and after hop in view,
New summaries bloom fresh and new.
The history trail is clear—
I twitch my nose and cheer!

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 36.44% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title matches the Activity view redesign and highlights responsive layout, error containment, and review hardening.
Description check	✅ Passed	The description is clearly about the Activity view rework and its follow-on review/polish changes.
Linked Issues check	✅ Passed	The changes implement the requested before/after activity previews and historical reconstruction instead of comparing against the current version.
Out of Scope Changes check	✅ Passed	The rest of the changes are supporting work for the Activity redesign, preview rendering, tests, and generated types; no clear unrelated scope stands out.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch claude/activity-redesign-review-vpaqmr

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[myieye]

Merge interactions with neighboring open PRs (for whoever merges second):

• Add variant model, CRDT changes, FwData bridge and sync (1/3) #2416 (variants model, 1/3) adds new CRDT change types. The coverage tests in this PR are deliberately hostile to that: ChangeTypes.ts regenerates on build, and the frontend tests fail until every new $type has a summary handler (or an explicit intentionally-generic allowance) and every new create type has a bulk-collapse noun. That's a ~15-minute, well-signposted task in change-summary.ts — the test failure messages say what to add. Expect it whenever Add variant model, CRDT changes, FwData bridge and sync (1/3) #2416 and this are both in.
• New Crowdin updates #2355 (Crowdin l10n_develop) will conflict in all 8 .po files — this PR adds ~96 msgids plus translator-context comments. Resolution: merge develop into the Crowdin branch, run pnpm -C frontend/viewer run i18n:extract (it restores comments), and let Crowdin re-sync. The /crowdin-merge flow already does this.
• Display pictures in senses #2412 (pictures in senses): the sensePicture facts and CreateSensePictureChange handlers are already covered here (they came in via develop merges). If Display pictures in senses #2412 adds further change types, same coverage-test friction as Add variant model, CRDT changes, FwData bridge and sync (1/3) #2416.
• Media manager #2358 (media manager): activity summaries currently assume every RemoteResource is audio (documented in describeChange — "the only resource kind today"). When non-audio media ships, that assumption needs a mime-based kind; the comment marks the spot.
• Rework the activity view: readable summaries and a before/after diff #2388: superseded by this PR (same branch history + review pass); merging both would double-land the feature — pick one.

Landmine note: HistoryService.ProjectActivity changed to Task<ProjectActivity[]>; any in-flight branch calling the old IAsyncEnumerable shape (e.g. new tests written against it) gets a compile error with an obvious fix — drop the .ToArrayAsync().

---

Generated by Claude Code

[argos-ci]

The latest updates on your projects. Learn more about Argos notifications ↗︎

Build	Status	Details	Updated (UTC)
default (Inspect)	⚠️ Changes detected (Review)	4 changed	Jul 12, 2026, 12:49 PM
e2e (Inspect)	⚠️ Changes detected (Review)	7 changed	Jul 12, 2026, 12:57 PM

[myieye]

@coderabbitai review

---

Generated by Claude Code

[coderabbitai]

The PR description image (mobile list) shows the activity feed in the narrow/mobile layout: a full-width list of activity cards with filter dropdowns ("All authors", "All activity types") and a sort control ("Newest first") at the top. Cards show entries like "Deleted word Apfel", grouped changes for "Created word wulu₂" (with nested "Added meaning a flying insect", "Added publication Main Dictionary"), homograph note "wulu₁ Set homograph number to 1", and other field-edit rows, each with an "Unknown" author and relative timestamp.

The PR description image (mobile detail) shows the full-screen detail view opened after tapping a row: a close (X) button, author/timestamp header, "Not synced" badge, a "History" button, Preview/Details tabs, and a red-washed "Deleted word Apfel" panel with fields like Word, Display as, Uses components as, Literal meaning, Note, etc.

Thanks for the context on merge interactions with #2416, #2355, #2412, #2358, and #2388 — noted for when those land alongside this one. Kicking off the review now.

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

Actionable comments posted: 6

🧹 Nitpick comments (5)

frontend/viewer/src/lib/activity/assemble-entry.test.ts (1)

17-28: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Consider adding a test for the "latest wins" deduplication behavior.

The latestById helper takes the latest snapshot when an entity id appears multiple times in the commit, but no test exercises this path. A duplicate-sense test would guard against regressions in the deduplication logic.

🧪 Suggested test

   it('ignores senses that belong to a different entry', () => {
     const contexts = [ctx('Entry', entrySnap), ctx('Sense', {id: 's1', entryId: 'other'})];
     expect(assembleEntryAtCommit(entryId, contexts, [])?.senses).toEqual([]);
   });
+
+  it('uses the latest snapshot when a sense appears multiple times in the commit', () => {
+    const contexts = [
+      ctx('Entry', entrySnap),
+      ctx('Sense', {id: 's1', entryId, gloss: {en: 'old gloss'}}),
+      ctx('Sense', {id: 's1', entryId, gloss: {en: 'new gloss'}}),
+    ];
+    const entry = assembleEntryAtCommit(entryId, contexts, []);
+    expect(entry?.senses).toHaveLength(1);
+    expect(entry?.senses[0].gloss).toEqual({en: 'new gloss'});
+  });
 });

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/lib/activity/assemble-entry.test.ts` around lines 17 -
28, The test covering assembleEntryAtCommit should also exercise latestById
deduplication by including duplicate Sense contexts with the same id and
different snapshots. Assert that the assembled entry contains one instance of
that sense using the latest snapshot’s values, while preserving the existing
commit-order and example-sentence assertions.

frontend/viewer/src/lib/entry-editor/diff-view/vocab-diff-coverage.test.ts (1)

13-26: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Prefer resolving generated paths relative to the test file, not process.cwd().

The sibling coverage test (entity-config-coverage.test.ts) resolves the same generated directory via fileURLToPath(import.meta.url), which is independent of the working directory. Using process.cwd() here couples the test to being run from frontend/viewer; aligning both tests removes that hidden assumption.

♻️ Suggested alignment

-import {describe, it, expect} from 'vitest';
-import {readFileSync} from 'node:fs';
-import {resolve} from 'node:path';
+import {describe, it, expect} from 'vitest';
+import {readFileSync} from 'node:fs';
+import {fileURLToPath} from 'node:url';
+import {dirname, join} from 'node:path';
 import {IGNORED_VOCAB_FIELDS} from './vocab-diff-fields';
+
+const here = dirname(fileURLToPath(import.meta.url));

Then derive the model/component paths from here instead of process.cwd().

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/lib/entry-editor/diff-view/vocab-diff-coverage.test.ts`
around lines 13 - 26, Update propertyNames and handledFields to resolve paths
from the test file location using the existing
fileURLToPath(import.meta.url)-based here pattern from
entity-config-coverage.test.ts, rather than process.cwd(). Derive both generated
model and DiffVocabPrimitive.svelte paths relative to here while preserving the
current file parsing behavior.

frontend/viewer/src/lib/activity/change-summary.ts (1)

98-104: 🎯 Functional Correctness | 🔵 Trivial | 💤 Low value

?? fallback conflates a camelCase null with an absent value.

prop returns record[camel] ?? record[pascal], so a legitimately-null camelCase property falls through to the (usually undefined) PascalCase variant. Today this is harmless — the only null-sensitive caller (SetPartOfSpeechChange) treats null and undefined identically — but any future handler that needs to distinguish "explicitly cleared to null" from "absent" would silently break. If you want to future-proof, prefer key presence over ??.

♻️ Presence-based lookup

-  const camel = key[0].toLowerCase() + key.slice(1);
-  const pascal = key[0].toUpperCase() + key.slice(1);
-  return record[camel] ?? record[pascal];
+  const camel = key[0].toLowerCase() + key.slice(1);
+  const pascal = key[0].toUpperCase() + key.slice(1);
+  return camel in record ? record[camel] : record[pascal];

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/lib/activity/change-summary.ts` around lines 98 - 104,
Update the prop function’s camelCase/PascalCase lookup to use property presence
rather than nullish coalescing, so an explicitly null camelCase value is
returned as null while the PascalCase variant is used only when the camelCase
key is absent.

frontend/viewer/src/lib/activity/ChangeSummary.svelte (1)

220-222: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Split the dense editObjectField one-liner for readability.

Line 222 nests object-type label, subject token, cleared/ws/value branches, and two more nested conditionals all on a single line. Since whitespace at {#if}/{:else}/snippet boundaries is trimmed (per this file's own documented spacing contract), reformatting across multiple lines is behavior-preserving and would make this branch much easier to scan/debug.

♻️ Suggested reformat (behavior-preserving)

-{:else if fact.kind === 'editObjectField'}
-  {`@const` objectType = capitalize(objectNoun(fact.object))}
-  {objectType}{`#if` subject}{`@render` subjectToken(subject, true)}{/if}{`#if` fact.cleared}{$t`Cleared ${fact.field}`}{`#if` fact.ws}{`@render` wsCode(fact.ws)}{/if}{:else if fact.value !== undefined}{`@const` value = fact.value}{`#if` fact.ws}<T msg="Set # to #" cmt="First # is a field label plus writing-system code, second # is the new value">{fact.field}{`@render` wsCode(fact.ws)}{`#snippet` second()}{`@render` chip(value, true)}{/snippet}</T>{:else}{$t`Set ${fact.field} to`}{`@render` chip(value)}{/if}{:else}{$t`Changed ${fact.field}`}{/if}
+{:else if fact.kind === 'editObjectField'}
+  {`@const` objectType = capitalize(objectNoun(fact.object))}
+  {objectType}
+  {`#if` subject}{`@render` subjectToken(subject, true)}{/if}
+  {`#if` fact.cleared}
+    {$t`Cleared ${fact.field}`}{`#if` fact.ws}{`@render` wsCode(fact.ws)}{/if}
+  {:else if fact.value !== undefined}
+    {`@const` value = fact.value}
+    {`#if` fact.ws}
+      <T msg="Set # to #" cmt="First # is a field label plus writing-system code, second # is the new value">{fact.field}{`@render` wsCode(fact.ws)}{`#snippet` second()}{`@render` chip(value, true)}{/snippet}</T>
+    {:else}
+      {$t`Set ${fact.field} to`}{`@render` chip(value)}
+    {/if}
+  {:else}
+    {$t`Changed ${fact.field}`}
+  {/if}

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/lib/activity/ChangeSummary.svelte` around lines 220 -
222, Reformat the `editObjectField` branch in `ChangeSummary.svelte` into
readable multiline markup, separating the `objectType`/`subjectToken` output
from the `cleared`, `value`, writing-system, and fallback conditionals. Preserve
the existing rendering order, conditions, translations, snippets, and whitespace
behavior; make no functional changes.

backend/FwLite/LcmCrdt/HistoryService.cs (1)

336-348: 🚀 Performance & Scalability | 🔵 Trivial | ⚡ Quick win

LoadPreviousSnapshot materializes the entity's full commit history to find one adjacent commit.

affectingCommitIds loads every commit ID that ever touched the entity, then does an in-memory IndexOf to locate the adjacent one. For heavily-edited entries this is an unbounded list load on every "view change details" call, where a targeted query (e.g., filter by ordering key relative to commitId, OrderByDescending().Take(1)) would return the same result with a bounded query.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@backend/FwLite/LcmCrdt/HistoryService.cs` around lines 336 - 348, Update
LoadPreviousSnapshot to query only the single most recent commit affecting
entityId that precedes commitId, using the existing commit ordering and a
descending filter with Take(1). Remove the full affectingCommitIds
materialization and in-memory IndexOf logic, preserve the null result when no
earlier affecting commit exists, and pass the selected commit ID to
dataModel.GetAtCommit.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@backend/FwLite/LcmCrdt/ActivityChangeInfoResolver.cs`:
- Around line 254-269: Update RecoverDeleted to select only the latest snapshot
per missing EntityId within the database query, rather than materializing all
snapshots and grouping with group.Last() in memory. Apply the same query-level
latest-row filtering in LoadComponentEndpointsFromCreate for
ChangeEntity<IChange>, before narrowing results to AddEntryComponentChange, and
remove the redundant in-memory grouping.

In `@backend/FwLite/LcmCrdt/HistoryService.cs`:
- Around line 57-59: Make ProjectActivity immutable by changing ChangeInfo from
a mutable setter to an init-only property, then update
ActivityChangeInfoResolver.ResolveAsync to return new ProjectActivity instances
using with expressions populated with the resolved ChangeInfo instead of
mutating existing activities or their arrays. Update callers to use the returned
activities while preserving the existing ordering and enrichment behavior.
- Around line 324-334: Update the previous-snapshot lookup in the change-context
construction to use the selected change’s identity, including its changeIndex,
rather than only commitId and entityId. Adjust LoadPreviousSnapshot and its
caller as needed so later same-entity changes in an AddChanges commit compare
against the state immediately before that specific change; keep snapshot
resolution and affected-entry handling unchanged.

In `@frontend/viewer/src/lib/history/HistoryView.svelte`:
- Line 106: Update the ActivityItem invocation in HistoryView to pass
record.changeInfo for the changeInfo field instead of an empty array, while
preserving the existing record change and changeTypes values.

In `@frontend/viewer/src/locales/en.po`:
- Around line 222-224: Rerun the locale catalog extraction to regenerate
frontend/viewer/src/locales/en.po, ensuring the {0} activity-log translator note
attaches to the intended activity-log string rather than “Add your first entry
to start building your dictionary.” and the Code field-label note attaches to
“Code” rather than “Closed.”

In `@frontend/viewer/src/locales/es.po`:
- Around line 227-230: Move the extractor comments from the generated es.po
entry to their corresponding source strings in DashboardView.svelte: place the
fallback activity-log comment on the matching list-field fallback string and the
translation-context comment on “Total entries.” Remove the misleading comments
from the empty-state prompt and Total entries entries in the generated PO
output.

---

Nitpick comments:
In `@backend/FwLite/LcmCrdt/HistoryService.cs`:
- Around line 336-348: Update LoadPreviousSnapshot to query only the single most
recent commit affecting entityId that precedes commitId, using the existing
commit ordering and a descending filter with Take(1). Remove the full
affectingCommitIds materialization and in-memory IndexOf logic, preserve the
null result when no earlier affecting commit exists, and pass the selected
commit ID to dataModel.GetAtCommit.

In `@frontend/viewer/src/lib/activity/assemble-entry.test.ts`:
- Around line 17-28: The test covering assembleEntryAtCommit should also
exercise latestById deduplication by including duplicate Sense contexts with the
same id and different snapshots. Assert that the assembled entry contains one
instance of that sense using the latest snapshot’s values, while preserving the
existing commit-order and example-sentence assertions.

In `@frontend/viewer/src/lib/activity/change-summary.ts`:
- Around line 98-104: Update the prop function’s camelCase/PascalCase lookup to
use property presence rather than nullish coalescing, so an explicitly null
camelCase value is returned as null while the PascalCase variant is used only
when the camelCase key is absent.

In `@frontend/viewer/src/lib/activity/ChangeSummary.svelte`:
- Around line 220-222: Reformat the `editObjectField` branch in
`ChangeSummary.svelte` into readable multiline markup, separating the
`objectType`/`subjectToken` output from the `cleared`, `value`, writing-system,
and fallback conditionals. Preserve the existing rendering order, conditions,
translations, snippets, and whitespace behavior; make no functional changes.

In `@frontend/viewer/src/lib/entry-editor/diff-view/vocab-diff-coverage.test.ts`:
- Around line 13-26: Update propertyNames and handledFields to resolve paths
from the test file location using the existing
fileURLToPath(import.meta.url)-based here pattern from
entity-config-coverage.test.ts, rather than process.cwd(). Derive both generated
model and DiffVocabPrimitive.svelte paths relative to here while preserving the
current file parsing behavior.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 78fc9bdc-1307-4fe4-80c9-7ff6cd7629c5

📥 Commits

Reviewing files that changed from the base of the PR and between b5e8ecb and f1325d8.

⛔ Files ignored due to path filters (1)

• frontend/pnpm-lock.yaml is excluded by !**/pnpm-lock.yaml

📒 Files selected for processing (75)

• backend/FwLite/FwLiteShared/Services/HistoryServiceJsInvokable.cs
• backend/FwLite/FwLiteShared/TypeGen/ChangeTypesCodeGenerator.cs
• backend/FwLite/FwLiteShared/TypeGen/ChangeTypesMarker.cs
• backend/FwLite/FwLiteShared/TypeGen/ReinforcedFwLiteTypingConfig.cs
• backend/FwLite/LcmCrdt.Tests/DateTimeOffsetOrmParityTests.cs
• backend/FwLite/LcmCrdt.Tests/HistoryServiceActivityTests.cs
• backend/FwLite/LcmCrdt.Tests/OpenProjectTests.cs
• backend/FwLite/LcmCrdt/ActivityChangeInfoResolver.cs
• backend/FwLite/LcmCrdt/HistoryService.cs
• frontend/viewer/package.json
• frontend/viewer/src/lib/activity/ActivityFilter.svelte
• frontend/viewer/src/lib/activity/ActivityItem.svelte
• frontend/viewer/src/lib/activity/ActivityItemChangePreview.svelte
• frontend/viewer/src/lib/activity/ActivityView.svelte
• frontend/viewer/src/lib/activity/ActivityViewPicker.svelte
• frontend/viewer/src/lib/activity/AuthorLabel.svelte
• frontend/viewer/src/lib/activity/ChangeSummary.svelte
• frontend/viewer/src/lib/activity/CollapsedEntryDiff.svelte
• frontend/viewer/src/lib/activity/CollapsedSenseDiff.svelte
• frontend/viewer/src/lib/activity/EntryHeadwordButton.svelte
• frontend/viewer/src/lib/activity/PreviewViewScope.svelte
• frontend/viewer/src/lib/activity/assemble-entry.test.ts
• frontend/viewer/src/lib/activity/assemble-entry.ts
• frontend/viewer/src/lib/activity/author-color-service.svelte.ts
• frontend/viewer/src/lib/activity/author-color.test.ts
• frontend/viewer/src/lib/activity/author-color.ts
• frontend/viewer/src/lib/activity/change-summary.test.ts
• frontend/viewer/src/lib/activity/change-summary.ts
• frontend/viewer/src/lib/activity/fact-glyph.ts
• frontend/viewer/src/lib/activity/preview-entity-coverage.test.ts
• frontend/viewer/src/lib/components/ListItem.svelte
• frontend/viewer/src/lib/components/dictionary/DictionaryEntry.svelte
• frontend/viewer/src/lib/components/dictionary/Headwords.svelte
• frontend/viewer/src/lib/components/field-editors/multi-ws-input.svelte
• frontend/viewer/src/lib/components/field-editors/rich-multi-ws-input.svelte
• frontend/viewer/src/lib/components/writing-system/WsCode.svelte
• frontend/viewer/src/lib/dotnet-types/generated-types/LcmCrdt/ChangeTypes.ts
• frontend/viewer/src/lib/dotnet-types/generated-types/LcmCrdt/IActivityChangeInfo.ts
• frontend/viewer/src/lib/dotnet-types/generated-types/LcmCrdt/IChangeContext.ts
• frontend/viewer/src/lib/dotnet-types/generated-types/LcmCrdt/IProjectActivity.ts
• frontend/viewer/src/lib/dotnet-types/generated-types/LcmCrdt/index.ts
• frontend/viewer/src/lib/entry-editor/diff-view/DiffAudio.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffEntryLinkList.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffEntryPrimitive.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffExamplePrimitive.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffMultiSelect.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffMultiString.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffRichText.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffSelect.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffSensePrimitive.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffShell.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffText.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/DiffVocabPrimitive.svelte
• frontend/viewer/src/lib/entry-editor/diff-view/diff-classes.ts
• frontend/viewer/src/lib/entry-editor/diff-view/diff-field-coverage.test.ts
• frontend/viewer/src/lib/entry-editor/diff-view/inline-diff.test.ts
• frontend/viewer/src/lib/entry-editor/diff-view/inline-diff.ts
• frontend/viewer/src/lib/entry-editor/diff-view/vocab-diff-coverage.test.ts
• frontend/viewer/src/lib/entry-editor/diff-view/vocab-diff-fields.ts
• frontend/viewer/src/lib/history/HistoryView.svelte
• frontend/viewer/src/lib/layout/ListDetailLayout.svelte
• frontend/viewer/src/lib/utils/search-params.ts
• frontend/viewer/src/lib/views/entity-config-coverage.test.ts
• frontend/viewer/src/locales/en.po
• frontend/viewer/src/locales/es.po
• frontend/viewer/src/locales/fr.po
• frontend/viewer/src/locales/id.po
• frontend/viewer/src/locales/ko.po
• frontend/viewer/src/locales/ms.po
• frontend/viewer/src/locales/sw.po
• frontend/viewer/src/locales/vi.po
• frontend/viewer/src/project/browse/BrowseView.svelte
• frontend/viewer/src/stories/activity/previews.stories.svelte
• frontend/viewer/src/stories/activity/summaries.stories.svelte
• frontend/viewer/src/stories/editor/diff/diff-overview.stories.svelte

[coderabbitai]

🚀 Performance & Scalability | 🟠 Major | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Look for any existing "latest snapshot per entity" helper in SIL.Harmony usage across the backend,
# which could be reused here instead of loading full history.
rg -n -C3 --type=cs 'Snapshots\.' backend/FwLite | rg -v 'ActivityChangeInfoResolver'
rg -n -C3 --type=cs 'DefaultOrder\(' backend/FwLite

Repository: sillsdev/languageforge-lexbox

Length of output: 6329

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Inspect the resolver around the two relevant methods.
FILE=backend/FwLite/LcmCrdt/ActivityChangeInfoResolver.cs
wc -l "$FILE"
sed -n '240,350p' "$FILE"

# Inspect the related history query for any reusable patterns.
FILE=backend/FwLite/LcmCrdt/HistoryService.cs
sed -n '273,320p' "$FILE"

Repository: sillsdev/languageforge-lexbox

Length of output: 8157

---

RecoverDeleted and LoadComponentEndpointsFromCreate should filter earlier

• RecoverDeleted loads every snapshot for each missing id and then does GroupBy(...).Last() in memory, even though only the latest snapshot is needed.
• LoadComponentEndpointsFromCreate does the same for ChangeEntity<IChange> before narrowing to AddEntryComponentChange.
• Push the “latest/create row only” selection into the query to avoid avoidable I/O and memory pressure as history grows.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@backend/FwLite/LcmCrdt/ActivityChangeInfoResolver.cs` around lines 254 - 269,
Update RecoverDeleted to select only the latest snapshot per missing EntityId
within the database query, rather than materializing all snapshots and grouping
with group.Last() in memory. Apply the same query-level latest-row filtering in
LoadComponentEndpointsFromCreate for ChangeEntity<IChange>, before narrowing
results to AddEntryComponentChange, and remove the redundant in-memory grouping.

Source: Coding guidelines

[coderabbitai]

📐 Maintainability & Code Quality | 🟠 Major | ⚡ Quick win

Mutable setter on ProjectActivity breaks record immutability.

ChangeInfo uses { get; set; } and is mutated post-construction by the resolver, which conflicts with the repo guideline to prefer immutable records for DTOs. Consider having ActivityChangeInfoResolver.ResolveAsync return the resolved list (or use with expressions) instead of mutating the array elements in place.

As per coding guidelines: "Prefer records for DTOs and immutable data structures".

♻️ Suggested direction

-    /// <summary>Resolved display info per change, parallel to <see cref="Changes"/> by index. Set during enrichment.</summary>
-    public IReadOnlyList<ActivityChangeInfo> ChangeInfo { get; set; } = [];
+    /// <summary>Resolved display info per change, parallel to <see cref="Changes"/> by index. Set during enrichment.</summary>
+    public IReadOnlyList<ActivityChangeInfo> ChangeInfo { get; init; } = [];

This would require ActivityChangeInfoResolver.ResolveAsync to return new ProjectActivity instances (e.g. via activity with { ChangeInfo = ... }) rather than mutating the passed-in array in place.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@backend/FwLite/LcmCrdt/HistoryService.cs` around lines 57 - 59, Make
ProjectActivity immutable by changing ChangeInfo from a mutable setter to an
init-only property, then update ActivityChangeInfoResolver.ResolveAsync to
return new ProjectActivity instances using with expressions populated with the
resolved ChangeInfo instead of mutating existing activities or their arrays.
Update callers to use the returned activities while preserving the existing
ordering and enrichment behavior.

Source: Coding guidelines

[coderabbitai]

🎯 Functional Correctness | 🟠 Major | 🏗️ Heavy lift

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Check whether commits can bundle multiple ChangeEntities against the same entity,
# and whether existing tests cover per-changeIndex before/after within a multi-change commit.
rg -n 'ChangeEntities' backend/FwLite/LcmCrdt -g '*.cs' -B2 -A8
fd -HI 'HistoryServiceActivityTests.cs' backend/FwLite/LcmCrdt.Tests -x cat {}

Repository: sillsdev/languageforge-lexbox

Length of output: 50386

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Inspect the relevant methods and surrounding history logic.
sed -n '300,380p' backend/FwLite/LcmCrdt/HistoryService.cs

# Look for any per-changeIndex snapshot handling or same-entity multi-change handling.
rg -n 'changeIndex|LoadPreviousSnapshot|GetAtCommit<|ChangeEntities.*Index|Index == changeIndex|previousSnapshot' backend/FwLite/LcmCrdt -g '*.cs' -A3 -B3

# Inspect the commit/change entity model to see whether multiple changes per entity in one commit are represented anywhere.
rg -n 'class ChangeEntity|record ChangeEntity|EntityId|Index' backend/FwLite/LcmCrdt -g '*.cs' -A4 -B4

Repository: sillsdev/languageforge-lexbox

Length of output: 50386

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo '--- HistoryService.cs (relevant section) ---'
sed -n '300,370p' backend/FwLite/LcmCrdt/HistoryService.cs

echo
echo '--- LoadPreviousSnapshot definition ---'
rg -n 'LoadPreviousSnapshot\(' backend/FwLite/LcmCrdt/HistoryService.cs -A30 -B5

echo
echo '--- Activity/history tests mentioning changeIndex, previousSnapshot, or multiple changes ---'
rg -n 'changeIndex|previousSnapshot|LoadPreviousSnapshot|GetChangeContext|Activity' backend/FwLite/LcmCrdt.Tests -g '*Tests.cs' -A4 -B4 --max-count 80

Repository: sillsdev/languageforge-lexbox

Length of output: 50385

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo '--- Search for multi-change commit creation helpers and overloads ---'
rg -n 'AddChange\(|AddChanges\(|ChangesNameHelper|CreateCommit|CommitMetadata' backend/FwLite/LcmCrdt -g '*.cs' -A4 -B4 --max-count 120

echo
echo '--- Inspect HistoryService change-context callers/usages ---'
rg -n 'LoadChangeContext\(|ChangeContext\(' backend/FwLite/LcmCrdt -g '*.cs' -A4 -B4 --max-count 60

echo
echo '--- Inspect HistoryServiceActivityTests for any commit with multiple changes in one commit ---'
rg -n 'AddChange\(ClientId, .*Meta\(\)\)|AddChange\(ClientId, .*CommitMetadata|await DataModel.AddChange' backend/FwLite/LcmCrdt.Tests/HistoryServiceActivityTests.cs -A3 -B3 --max-count 120

Repository: sillsdev/languageforge-lexbox

Length of output: 50385

---

Compute previousSnapshot from the selected change, not just the commit
LoadPreviousSnapshot(commitId, entityId) returns the state before the last commit that touched the entity. For commits built with AddChanges(...), a later changeIndex on the same entity will compare against the pre-commit state instead of the state before that specific change, so the Activity diff can attribute earlier same-commit edits to the wrong change.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@backend/FwLite/LcmCrdt/HistoryService.cs` around lines 324 - 334, Update the
previous-snapshot lookup in the change-context construction to use the selected
change’s identity, including its changeIndex, rather than only commitId and
entityId. Adjust LoadPreviousSnapshot and its caller as needed so later
same-entity changes in an AddChanges commit compare against the state
immediately before that specific change; keep snapshot resolution and
affected-entry handling unchanged.

[coderabbitai]

🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Inspect how ActivityItem.svelte consumes activity.changes vs activity.changeInfo
fd -a 'ActivityItem.svelte' | xargs -I{} sh -c 'echo "== {} =="; cat -n "{}"'

Repository: sillsdev/languageforge-lexbox

Length of output: 21074

---

Pass the record’s changeInfo through here
ActivityItem looks up changeInfo by change.change.index, so hardcoding [] drops the subject/root metadata for every rendered change. Pass the existing record.changeInfo instead.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/lib/history/HistoryView.svelte` at line 106, Update the
ActivityItem invocation in HistoryView to pass record.changeInfo for the
changeInfo field instead of an empty array, while preserving the existing record
change and changeTypes values.

[coderabbitai]

📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Flag PO entries whose #. comment references {0}/{1} but the msgid has no placeholder,
# which indicates a comment attached to the wrong msgid.
fd -e po . frontend/viewer/src/locales --exec python3 - {} <<'PY'
import sys, re
path = sys.argv[1]
block = []
def check(block, path):
    comments = [l for l in block if l.startswith('#.')]
    msgid = next((l for l in block if l.startswith('msgid ')), '')
    ctext = ' '.join(comments)
    if re.search(r'\{[01]\}', ctext) and not re.search(r'\{[01]\}|#', msgid):
        print(f"{path}: comment mentions placeholder but msgid has none -> {msgid.strip()}")
with open(path, encoding='utf-8') as f:
    for line in f:
        if line.strip() == '':
            check(block, path); block = []
        else:
            block.append(line)
    check(block, path)
PY

Repository: sillsdev/languageforge-lexbox

Length of output: 167

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

file='frontend/viewer/src/locales/en.po'

python3 - <<'PY'
from pathlib import Path
import re

path = Path('frontend/viewer/src/locales/en.po')
lines = path.read_text(encoding='utf-8').splitlines()

targets = [33, 222, 535]
for t in targets:
    start = max(1, t - 6)
    end = min(len(lines), t + 10)
    print(f"\n=== lines {start}-{end} around {t} ===")
    for i in range(start, end + 1):
        print(f"{i:4}: {lines[i-1]}")
PY

Repository: sillsdev/languageforge-lexbox

Length of output: 2757

---

🏁 Script executed:

python3 - <<'PY'
from pathlib import Path

path = Path('frontend/viewer/src/locales/en.po')
lines = path.read_text(encoding='utf-8').splitlines()

start, end = 545, 556
for i in range(start, end + 1):
    print(f"{i:4}: {lines[i-1]}")
PY

Repository: sillsdev/languageforge-lexbox

Length of output: 623

---

🏁 Script executed:

python3 - <<'PY'
from pathlib import Path
import re

path = Path('frontend/viewer/src/locales/en.po')
lines = path.read_text(encoding='utf-8').splitlines()

for needle in ['msgid "Code"', 'msgid "Closed"', 'msgid "(unknown)"', 'msgid "Add your first entry to start building your dictionary."']:
    print(f"\n=== {needle} ===")
    for i, line in enumerate(lines, 1):
        if line == needle:
            start = max(1, i-3)
            end = min(len(lines), i+3)
            for j in range(start, end+1):
                print(f"{j:4}: {lines[j-1]}")
            print()
PY

Repository: sillsdev/languageforge-lexbox

Length of output: 1869

---

Realign the misplaced translator comments

frontend/viewer/src/locales/en.po:222-224, 535-537 has two stale #. comments attached to the wrong msgid:

• the {0} activity-log context sits above Add your first entry to start building your dictionary.
• the Code field-label context sits above Closed

Rerun the catalog extraction so the translator notes line up with the intended strings.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/locales/en.po` around lines 222 - 224, Rerun the locale
catalog extraction to regenerate frontend/viewer/src/locales/en.po, ensuring the
{0} activity-log translator note attaches to the intended activity-log string
rather than “Add your first entry to start building your dictionary.” and the
Code field-label note attaches to “Code” rather than “Closed.”

[coderabbitai]

📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Locate the source annotations to confirm the misattributed extractor comments.
rg -nP -C3 'Add your first entry to start building your dictionary' -g '*.svelte'
rg -nP -C3 'fallback for list fields without a singular noun' -g '*.svelte' -g '*.ts'
rg -nP -C3 'Total entries' -g '*.svelte'

Repository: sillsdev/languageforge-lexbox

Length of output: 167

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo '--- es.po context around the flagged entry ---'
python3 - <<'PY'
from pathlib import Path
p = Path('frontend/viewer/src/locales/es.po')
lines = p.read_text(encoding='utf-8').splitlines()
for start in [220, 2668]:
    print(f'\n### {p}:{start}-{start+20}')
    for i in range(start-1, min(len(lines), start+20)):
        print(f'{i+1}:{lines[i]}')
PY

echo
echo '--- search for the exact msgids and nearby source comments ---'
rg -n -C 2 -F 'Add your first entry to start building your dictionary.' .
rg -n -C 2 -F 'Total entries' .
rg -n -C 2 -F 'fallback for list fields without a singular noun' .
rg -n -C 2 -F 'A chip with the added item follows.' .
rg -n -C 2 -F 'translation' frontend/viewer/src frontend .

Repository: sillsdev/languageforge-lexbox

Length of output: 50387

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo '--- inventory of locale-related source files ---'
git ls-files 'frontend/viewer/src/**/*' | rg 'locale|locales|svelte|ts|js|po|json'

echo
echo '--- candidate extractor/source annotation search across frontend/viewer/src ---'
rg -n -C 2 -F '{0} = plural field label' frontend/viewer/src || true
rg -n -C 2 -F 'plural field label' frontend/viewer/src || true
rg -n -C 2 -F 'A chip with the added item follows.' frontend/viewer/src || true
rg -n -C 2 -F 'Add your first entry to start building your dictionary.' frontend/viewer/src || true
rg -n -C 2 -F 'Total entries' frontend/viewer/src || true

Repository: sillsdev/languageforge-lexbox

Length of output: 50385

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo '--- inspect neighboring entries in frontend/viewer/src/locales/es.po ---'
python3 - <<'PY'
from pathlib import Path
p = Path('frontend/viewer/src/locales/es.po')
lines = p.read_text(encoding='utf-8').splitlines()
for start in [220, 2668]:
    print(f'\n### {p}:{start}-{start+18}')
    for i in range(start-1, min(len(lines), start+18)):
        print(f'{i+1}:{lines[i]}')
PY

Repository: sillsdev/languageforge-lexbox

Length of output: 1943

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo '--- DashboardView.svelte context around the dashboard prompt and totals ---'
ast-grep outline frontend/viewer/src/project/dashboard/DashboardView.svelte --view expanded || true
echo
rg -n -C 4 -F 'Add your first entry to start building your dictionary.' frontend/viewer/src/project/dashboard/DashboardView.svelte
echo
rg -n -C 4 -F 'Total entries' frontend/viewer/src/project/dashboard/DashboardView.svelte
echo
rg -n -C 4 -F 'A chip with the added item follows.' frontend/viewer/src/project/dashboard/DashboardView.svelte
echo
rg -n -C 4 -F 'fallback for list fields without a singular noun' frontend/viewer/src/project/dashboard/DashboardView.svelte

Repository: sillsdev/languageforge-lexbox

Length of output: 2403

---

Move the extractor comments to the matching strings in frontend/viewer/src/project/dashboard/DashboardView.svelte. The fallback comment is attached to the empty-state prompt, and the translation context comment is attached to Total entries; both are misleading in the generated .po.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@frontend/viewer/src/locales/es.po` around lines 227 - 230, Move the extractor
comments from the generated es.po entry to their corresponding source strings in
DashboardView.svelte: place the fallback activity-log comment on the matching
list-field fallback string and the translation-context comment on “Total
entries.” Remove the misleading comments from the empty-state prompt and Total
entries entries in the generated PO output.

Source: Learnings

[myieye]

Review absorbed in 89ef1e6. Fixed: init-only ChangeInfo with the resolver returning enriched copies; RecoverDeleted now takes one latest-snapshot query per missing id instead of loading full snapshot histories; component-endpoint recovery filters to AddEntryComponentChange in SQL; presence-based prop() lookup; the assemble-entry latest-wins test; import.meta.url paths in the vocab coverage test; and the five stale translator comments (they'd stuck to alphabetically-adjacent msgids and the comment-preserving extraction kept them alive by msgid — all 8 locales cleaned).

Skipped, with reasons:

• Per-changeIndex previous snapshots: Harmony stores one snapshot per entity per commit, so intra-commit intermediate states would require replaying changes — a heavy lift for a case the UI already handles differently (multi-change same-entity commits collapse to an assembled tree card rather than per-change diffs). Noted as a follow-up.
• record.changeInfo in HistoryView: HistoryLineItem has no changeInfo property — the suggestion doesn't compile. [] is the honest value until history lines get server-side resolution (follow-up).
• Bounded LoadPreviousSnapshot query: filtering commits relative to another commit's position means comparing HybridDateTime timestamps through linq2db, which this repo documents as unsafe (strftime millisecond truncation — see AGENTS.md). The ids-list shape deliberately avoids the comparison; it loads only GUIDs.
• ChangeSummary reformat: whitespace between inline expression nodes is not trimmed in Svelte (only block boundaries collapse) — the suggested multiline form would introduce stray spaces between the subject token and its margin-based spacing. The dense line is the file's documented spacing contract.

---

Generated by Claude Code

[myieye]

Closing in favour of the stacked merge sequence starting at #2435 — this PR was too large to merge in one piece. Its final head (89ef1e6, CodeRabbit review absorbed, all workflows green apart from Argos baseline approvals) is byte-identical to the tree you get after merging all six parts (claude/activity-1-backend-change-info → … → claude/activity-6-list-detail-layout), so this PR remains the reference for how the assembled feature looks and was reviewed. Only the step-1 PR is open; each next part gets its PR when its predecessor merges.

---

Generated by Claude Code