i18n(id): refresh Indonesian admin translations

[ahliweb]

What does this PR do?

Refreshes the Indonesian admin catalog. The fork main is synchronized with upstream emdash-cms/emdash:main at 450057e46e0f5cb6300e5cad127fb3b2281e1ad1.

Since the last translation-compatible base dd325cb1629f77fdac7f6494be8b752322812024, upstream has added Media Library detail/usage UI changes, delegated release-service and security hardening, revision/concurrency safeguards, UX acceptance infrastructure, sandboxed content:beforeSave rejection support, publish/autosave race hardening, release-service test stabilization, and multiple generated locale-catalog extractions.

Translation scope

The current PR head still preserves the previously reviewed Indonesian translations. Deterministic reconciliation against current main remains required because several upstream extractions happened after this branch was created.

Previously identified strings still requiring reconciliation include Media Library and delegated-release/verification strings such as:

• Move the focal point to choose what stays visible in cropped images.
• No usage
• No usage to show yet
• See where this file is used.
• This file isn’t used in any content.
• Authorized approvers
• Failed to verify plugin
• Independent verification
• Provenance optional
• Provenance required
• Publisher release policy
• Release CID
• Review Verified Plugin
• Review Verified Update
• Select a release before verifying it
• We couldn't verify this publisher's identity
• Plugin {pluginId} rejected the save: {reason}

Recommended Indonesian rendering for the newest identified message remains: Plugin {pluginId} menolak penyimpanan: {reason}.

Terminology refinements retained from the earlier review:

• destructive Discard actions use Buang / Membuang...;
• non-destructive Dismiss actions remain distinct;
• Heading terminology is normalized to Judul, including Judul 4–6;
• the technical UI term byline remains consistent with the existing Indonesian catalog.

Upstream watch notes

• fix(admin): save editor state before publishing emdash-cms/emdash#2860 (fix(admin): save editor state before publishing) merged as afa81c5e.... Root cause: Publish could race the autosave debounce/in-flight autosave and promote stale draft state, dropping the editor's latest changes or regressing already-live fields. The fix serializes save/autosave/publish, cancels pending debounce before publish, flushes the exact current payload, propagates revision tokens, and stops publish on validation/save/revision failure. Upstream reports 1,805 admin tests, build, workspace typecheck, lint, formatting, and git diff --check passing.
• fix(admin): save editor state before publishing emdash-cms/emdash#2860 explicitly adds no user-visible strings.
• 5cb3f73b... is the follow-up locale extraction. It changes generated source references across catalogs (79 additions / 79 deletions per catalog in the extraction commit) and does not introduce a new feature-level msgid from fix(admin): save editor state before publishing emdash-cms/emdash#2860. These generated references still need to be incorporated during deterministic PO reconciliation.
• test(release-service): isolate encryption verification sweep emdash-cms/emdash#2868 (test(release-service): isolate encryption verification sweep) merged as 450057e.... Root cause: the exhaustive 256-shard encryption verification sweep and time-sensitive publication-coordination expiry test were contending inside the main Workers-pool suite, causing repeated unrelated CI failures on upstream PRs. The fix isolates the exhaustive sweep in its own Vitest invocation and makes lease expiry deterministic without reducing coverage. It changes release-service tests/test configuration only and introduces no locale or admin UI strings. Scoped release-service tests, typecheck, and pnpm lint:quick passed upstream; the PR notes a pre-existing repository-wide type-aware lint baseline of 42 unrelated diagnostics.
• fix(admin): keep locale menus consistent with the admin theme emdash-cms/emdash#2870 (fix(admin): keep locale menus consistent with the admin theme) is open and currently mergeable. It replaces the native locale <select> with Kumo Select, preserves existing message IDs, and explicitly includes no messages.po changes. Root cause: the native menu did not inherit dark-mode/theming consistently and could interact poorly with unsaved byline state. If merged, it should not add translation debt beyond generated source-reference movement; upstream reports lint, typecheck, format check, 1,813 admin browser tests, focused locale/byline tests, and LTR/RTL visual checks passing on rerun.
• Add media image cropping emdash-cms/emdash#2861 (Add media image cropping) is open and currently mergeable. It adds substantial localized Media Details crop UI (focal/crop modes, aspect-ratio controls, create-copy/replace-original actions and confirmation text). Because user-visible strings are wrapped for translation and catalogs are intentionally extracted after merge, this PR is a likely source of a future Indonesian locale delta if it lands before i18n(id): refresh Indonesian admin translations #83 is reconciled. Upstream reports build, typecheck, lint, 6,262 core tests, 1,841 admin tests, focused Playwright coverage, and docs build passing. Its head advanced to 1a7b9f38... and changeset-bot now reports release changes across 17 packages; it remains open and mergeable.
• Issue Admin View published links ignore entry locale emdash-cms/emdash#2874 (Admin View published links ignore entry locale) is a newly reported i18n runtime bug: published-link construction ignores the entry locale while preview paths already support {locale}. It does not currently change the Indonesian catalog or PR i18n(id): refresh Indonesian admin translations #83 scope, but a future admin fix could move locale source references or add UI messaging.
• Issue Visual Portable Text publish leaves key-only pending changes emdash-cms/emdash#2877 (Visual Portable Text publish leaves key-only pending changes) is newly opened against 0.36.0/current behavior. Root cause reported by the reproducer: pmToPortableText() regenerates random _key values on repeated getBlocks() calls, while the visual-editor Portable Text save is not part of the toolbar's pendingSavePromise. A pagehide/keepalive flush can therefore persist a semantically identical post-publish draft that differs only by generated keys. This is a runtime autosave/publish consistency bug, not caused by the Indonesian catalog, and currently adds no locale delta.
• Issue Repeated (nested repeater sub-field) input reverts to a stale autosave payload if its response arrives after further edits emdash-cms/emdash#2878 (Repeated (nested repeater sub-field) input reverts to a stale autosave payload if its response arrives after further edits) is newly opened against 0.35.0. The reported root cause is a remaining stale-response race in patchAutosaveQueries: an earlier autosave response/request-time payload can patch observed query state after the user has continued editing a nested repeater field, reintroducing a narrower form of the stale autosave overwrite class previously addressed by Fix autosave form reset bug emdash-cms/emdash#302. This is also runtime state-management behavior outside PR i18n(id): refresh Indonesian admin translations #83's translation scope.
• Issues Allow trusted content-editor panels to read and patch the live draft emdash-cms/emdash#2866 and Add a Block Kit placement for sandboxed content-editor actions emdash-cms/emdash#2867 remain future extension-surface work for live-draft editor panels and sandboxed Block Kit editor actions; they do not currently modify this PR's branch.
• Clarify that Save creates a draft before publishing new content emdash-cms/emdash#2857 still proposes localized first-save guidance (Save draft or equivalent) but has not yet landed on main; it may create a future locale delta if merged.

Validation status

• fork main and upstream main are synchronized at 450057e46e0f5cb6300e5cad127fb3b2281e1ad1;
• current PR head remains b841cd537aa2f802360b2b63e0a879cab81183cd;
• comparison with current main: 1 commit ahead / 31 commits behind, merge base dd325cb1629f77fdac7f6494be8b752322812024;
• GitHub reports the PR as not mergeable;
• PR scope remains 1 commit / 2 files (.changeset/bright-id-refresh.md and packages/admin/src/locales/id/messages.po);
• the last validated head passed CI, Format, Query Counts, CodeQL, zizmor, Auto Format, and Visual Regression;
• Preview Releases on that old head remains the only failed workflow; refreshed-head validation is still required after reconciliation;
• no maintainer reviews, review requests, or unresolved review threads are present;
• Lunaria still warns that localization source changed after PR creation;
• contributor CLA remains incomplete and must be handled by the contributor; automation will not sign or accept it;
• a refreshed-head zero-gap locale validation and complete CI run are required after deterministic PO reconciliation/rebase.

Type of change

• Bug fix
• Feature
• Refactor
• Translation
• Documentation
• Performance improvement
• Tests
• Chore

Checklist

• I have read CONTRIBUTING.md
• latest upstream locale extraction reconciled
• zero-untranslated Indonesian catalog guard passes on refreshed head
• branch rebuilt atomically on current upstream main
• translation-only runtime impact; no application source behavior changed
• user-facing changeset added and reviewed
• refreshed-head CI, lint, formatting, type-checking, tests, build, and visual regression checks complete successfully
• contributor CLA requirement complete

AI-generated code disclosure

• This PR includes AI-assisted translation review and catalog maintenance — model/tool: GPT-5.6 Sol; no runtime code changes.

Screenshots / test output

No visual component/source changes are intended. This PR changes translation catalog content only. The branch must be deterministically refreshed against the latest upstream extraction before final merge-readiness validation.

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[github-actions]

Lunaria Status Overview

🌕 This pull request will trigger status changes.

Learn more

By default, every PR changing files present in the Lunaria configuration's files property will be considered and trigger status changes accordingly.

You can change this by adding one of the keywords present in the ignoreKeywords property in your Lunaria configuration file in the PR's title (ignoring all files) or by including a tracker directive in the merged commit's description.

Tracked Files

File	Note
packages/admin/src/locales/id/messages.po	Localization changed, will be marked as complete. 🔄️

Warnings reference

Icon	Description
🔄️	The source for this localization has been updated since the creation of this pull request, make sure all changes in the source have been applied.

[ahliweb]

Upstream watch update (2026-09-09): upstream emdash-cms/emdash:main advanced from locale extraction d6b99d1bd673c2772f4d69a51ebbb9be74461be1 to b2212bacd9cd5b9836e198e71fbb3a448065dfd5 via four non-locale commits. I fast-forwarded fork ahliweb/emdash-awcms:main atomically to the same SHA (no force update).

Post-extraction commits and root causes:

• fix(bot): recover workspace base after retries emdash-cms/emdash#2974 fix(bot): recover workspace base after retries — retry/resume could reconstruct ExecEnv without its process-local snapshot-base field even though the durable VFS marker still existed, causing writable commands to fail during checkpoint with workspace snapshot base is not initialized. The fix reloads the base ref from the marker before writable execution/checkpoint. Internal bot-only; no admin strings.
• fix: align OpenAPI content terms spec with shipped route emdash-cms/emdash#2925 fix(openapi): align content terms docs with shipped nested route — OpenAPI still documented a stale bare PUT .../terms route while the shipped API uses GET/POST .../terms/{taxonomy}. Documentation/schema correction only; no admin locale delta.
• fix(core): bound the media-usage cleanup scan with a row-value cursor emdash-cms/emdash#2910 fix(core): bound the media-usage cleanup scan with a row-value cursor — SQLite bound-parameter planning turned the OR-based keyset cursor into MULTI-INDEX OR + temp B-tree sorting, defeating LIMIT and scanning the backlog. Row-value comparison restores the composite-index range seek. Core/query-plan change only; no admin strings.
• ci: exclude generated changelogs from format checks emdash-cms/emdash#2976 ci: exclude generated changelogs from format checks — Changesets-generated CHANGELOG output could fail oxfmt in release PR ci: release emdash-cms/emdash#2833 even though source changesets are the maintained input. CI/tooling fix only; no locale delta.

The latest canonical Indonesian catalog extraction is therefore still d6b99d1...; there is no newer extraction after these four commits. PR #83 remains open/not mergeable at b841cd537aa2f802360b2b63e0a879cab81183cd. Against current fork main it is now 1 commit ahead / 110 commits behind, merge base dd325cb1629f77fdac7f6494be8b752322812024, with diff scope still exactly two files: .changeset/bright-id-refresh.md and packages/admin/src/locales/id/messages.po.

No submitted reviews, requested reviewers, or unresolved review threads are present. Historical-head Actions remain CI/Format/Query Counts/Visual Regression/CodeQL/zizmor/Auto Format green; Preview Releases is the old failure, specifically at pkg-pr-new publish after install and build had already succeeded.

The translation branch still needs deterministic msgid-based reconciliation onto current main using the d6b99d1... extracted catalog, followed by zero-empty/fuzzy, placeholder/ICU/tag parity, diff-scope, and refreshed-head CI/test/lint/typecheck/build/visual validation. I did not force-rewrite the large generated PO through the connector because that would not provide a safe local Lingui/catalog validation workspace. No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-09): upstream main advanced by one release commit to fc87efeb1df1d17e672e78af362628ebf7b9eddb (ci: release (#2833)). I fast-forwarded fork main to the same SHA without force. The release commit removes consumed changesets and updates package versions/changelogs; it does not modify packages/admin/src/locales/*/messages.po, so the canonical Indonesian catalog target remains the prior extraction (d6b99d1...) plus subsequent source changes already tracked. PR #83 is now 1 commit ahead / 111 commits behind current main; its diff scope remains exactly the changeset plus packages/admin/src/locales/id/messages.po.

A directly relevant upstream development is issue emdash-cms#2389 (production/preview builds rendering raw ICU when a feature branch introduces strings absent from the compiled catalog). Maintainer ascorbic instructed EmDashBot to make preview builds extract catalogs as part of the build. The bot candidate adds CI-only extraction before locale:compile, which addresses the root cause: pkg.pr.new builds currently compile stale .po files in production mode, where Lingui has no runtime compiler fallback. However, the bot could not run normal lint/typecheck/test/build verification because of its execution-environment restriction, could not include its changeset in the already-published candidate, and the candidate preview later timed out after 10 minutes. Therefore this is not yet a validated upstream fix and should not be copied into #83.

Upstream issue emdash-cms#2346 also received a bot candidate fixing user-linked byline translation inheritance (user_id) with 4 new regression tests and 89 focused byline tests passing, plus typecheck/lint/format. Its preview likewise timed out after 10 minutes. This is runtime locale behavior, not catalog content, so it does not change the Indonesian PO scope unless follow-up UI strings are added.

No maintainer review, requested reviewer, or unresolved review thread has appeared on #83. CLA remains contributor-owned and was not accepted or signed by automation.

[ahliweb]

Upstream watch update (2026-09-09): upstream emdash-cms/emdash:main and fork ahliweb/emdash-awcms:main remain synchronized at fc87efeb1df1d17e672e78af362628ebf7b9eddb; no new locale extraction has landed since the release commit.

New merge-horizon item: upstream PR emdash-cms#2979 (Improve passkey onboarding across account creation) opened against current main and is currently mergeable at a6a08e64de490b6f34f4a442edc1625eb5ae9767. It adds a substantial set of new Lingui-wrapped, user-visible passkey onboarding strings across invite acceptance, first-run setup, and signup (device/platform-specific explanations, Windows Hello/cross-device/security-key guidance, success-step copy, and even a newly localized Jane Doe placeholder). If emdash-cms#2979 lands before this Indonesian refresh is rebuilt, the post-merge extraction will create another real Indonesian catalog delta, not merely generated source-reference churn.

Root cause/context: current passkey account creation opens browser-owned WebAuthn UI with insufficient device-specific preparation. emdash-cms#2979 adds platform/capability detection and explanatory UI before invoking the credential prompt, following FIDO guidance. EmDashBot review found two implementation concerns: an ms-settings: Windows Settings LinkButton lacks the required external prop, and showSuccessStep is silently ignored unless showEducation is also true. The bot classifies neither as a blocker but explicitly says the LinkButton issue should be fixed before merge. Current checks are green for Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, and Auto Format; aggregate CI is still queued. The PR also carries a scope-guard warning (1,081 changed lines / 8 files), but remains mergeable.

PR #83 itself remains open/not mergeable at b841cd537aa2f802360b2b63e0a879cab81183cd, still 1 commit ahead / 111 commits behind current main, with the same two-file diff and merge base dd325cb1629f77fdac7f6494be8b752322812024. No review requests, submitted reviews, or unresolved review threads are present on #83. Historical head checks remain unchanged: all green except the old Preview Releases failure.

No branch mutation was performed in this check: the canonical Indonesian catalog still needs deterministic reconciliation against the latest extracted catalog, and emdash-cms#2979 has not merged/extracted yet. CLA remains contributor-owned; automation did not sign or accept any legal agreement.

[ahliweb]

Upstream watch update (2026-09-09): emdash-cms/emdash:main and fork ahliweb/emdash-awcms:main remain synchronized at fc87efeb1df1d17e672e78af362628ebf7b9eddb; there is still no locale extraction newer than the canonical d6b99d1... extraction already identified for deterministic Indonesian reconciliation.

Meaningful merge-horizon change: upstream emdash-cms#2979 (Improve passkey onboarding across account creation) advanced from its initial head to 11607eb08d63013b2479ccfa4c2bd5b020ed29bb after addressing EmDashBot review feedback. Commit 0eb2ca3e8... marks the Windows ms-settings:signinoptions LinkButton as external and makes the success confirmation respect showSuccessStep; both review threads are now resolved. Commit 11607eb... then fixes the E2E assumptions so setup/invite tests accept capability-dependent primary actions (Create passkey, Use another device, or Use a security key) instead of assuming a local authenticator is always available. Current-head CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, and Auto Format are all green.

emdash-cms#2979 still contains many Lingui-wrapped onboarding strings, so if it merges before #83 is rebuilt, the post-merge extraction will add further Indonesian translation work. No speculative PO edits were made because emdash-cms#2979 has not landed on main and therefore is not part of the canonical extracted catalog yet.

PR #83 itself remains open/not mergeable at b841cd537aa2f802360b2b63e0a879cab81183cd, with branch comparison still 1 commit ahead / 111 behind current main, merge base dd325cb1629f77fdac7f6494be8b752322812024, and scope limited to the changeset plus packages/admin/src/locales/id/messages.po. There are still no requested reviewers, submitted reviews, or review threads. No new upstream issues were opened or updated in this interval. CLA remains contributor-owned; no legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-09): emdash-cms/emdash:main advanced one commit from fc87efeb1df1d17e672e78af362628ebf7b9eddb to 4cc150e931313644a96b796627e5ec74b46c0aec via emdash-cms#2742 (feat(labeler): redesign operator review workflow). I fast-forwarded ahliweb/emdash-awcms:main to the same SHA without force; fork main is synchronized again.

This upstream commit is large (labeler/operator-review, registry moderation/lexicon and related release changes), but inspection of the commit diff found no packages/admin/src/locales/*/messages.po change and no new canonical locale extraction. Its admin package impact in the generated release PR is dependency-version propagation, not a new Indonesian catalog delta. Therefore no speculative PO rewrite was applied to i18n/id-refresh-2026-08-26.

After the fork sync, PR #83 remains on head b841cd537aa2f802360b2b63e0a879cab81183cd, still diverged from merge base dd325cb1629f77fdac7f6494be8b752322812024, now 1 commit ahead / 112 commits behind current main; GitHub still reports it not mergeable. Scope remains 1 commit / 2 files. There are still no requested reviewers, submitted reviews, or review threads. Old-head workflow state is unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor and Auto Format pass; Preview Releases is the lone historical failure; Dependabot Auto-Approve is skipped.

New upstream PR emdash-cms#2983 is the Changesets release PR created/updated from emdash-cms#2742 and mainly carries release/version propagation. PR emdash-cms#2980 (feat(plugins): add revision-based conditional writes to storage and KV) is also active, but explicitly has no admin UI changes and therefore no confirmed locale impact. No upstream issues were newly opened or updated in the checked interval, and no new Dependabot PR was found.

The translation branch still needs deterministic reconciliation against the latest canonical extracted catalog (d6b99d1… remains the last identified extraction point), preserving reviewed Indonesian msgstr, filling all new empty entries, validating zero untranslated/fuzzy entries plus ICU/placeholder/tag parity, confirming the two-file diff scope, and rerunning refreshed-head CI before merge-readiness can be reassessed.

CLA remains contributor-owned. No CLA or other legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-10): upstream emdash-cms/emdash:main advanced two commits from 4cc150e931313644a96b796627e5ec74b46c0aec to 88dc782acd849426187205450585d1f9f6caf784. I fast-forwarded fork ahliweb/emdash-awcms:main to the same SHA without force.

The two landed changes are emdash-cms#2985 (fix(registry): restore signed label projection) and emdash-cms#2984 (fix(labeler): keep review queue current). The delta does not touch packages/admin/src/locales/*/messages.po, so there is no new admin locale extraction to reconcile in this interval. emdash-cms#2985 fixes two production signed-label decoding/replay paths that could leave the fail-closed public projection empty; emdash-cms#2984 suppresses stale/replayed assessments from the current labeler review queue. Neither introduces Indonesian admin catalog debt.

New upstream issue emdash-cms#2988 reports that plugin admin pages declared with path: "/" never receive the sidebar active state. Root cause is a trailing-slash mismatch between the concatenated plugin target (/plugins/<id>/) and the router-normalized current path (/plugins/<id>). It currently has no locale/catalog change and is outside this translation PR's atomic scope.

After syncing main, i18n/id-refresh-2026-08-26 is still diverged at 1 commit ahead / 114 commits behind with merge base dd325cb1629f77fdac7f6494be8b752322812024; scope remains exactly the changeset plus packages/admin/src/locales/id/messages.po. PR #83 remains open/not mergeable, with no requested reviewers, submitted reviews, or review threads. Historical head workflows remain unchanged: CI/Format/Query Counts/Visual Regression/CodeQL/zizmor/Auto Format succeeded; Preview Releases is the old failure. No refreshed-head validation exists because the translation branch was not rewritten.

I did not modify the Indonesian branch because this interval added no canonical extraction, while its earlier deterministic catalog reconciliation is still outstanding. CLA remains contributor-owned and was not signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): emdash-cms/emdash:main advanced from 88dc782acd849426187205450585d1f9f6caf784 through six commits to canonical locale extraction 44114afd391ea0738bf95b4688d59513d2cb6347. I fast-forwarded ahliweb/emdash-awcms:main to that exact SHA without force; fork main is now synchronized.

The extraction is materially translation-relevant: packages/admin/src/locales/id/messages.po changed by 277 additions / 100 deletions. The principal source is merged emdash-cms#2979 (Improve passkey onboarding across account creation), which replaced/reworked passkey account-creation UX and added many Lingui-wrapped messages (device capability checks, Windows Hello / Touch ID guidance, another-device/security-key paths, confirmation/privacy guidance, and Create passkey). The extraction leaves those newly introduced messages untranslated in the canonical Indonesian catalog, so #83 now has additional real translation debt rather than reference churn only. The other intervening merges (emdash-cms#2992 labeler moderation, emdash-cms#2981 bot sandbox retry, emdash-cms#2970 edit-proxy preservation, emdash-cms#2967 workerd give-up diagnostics) do not themselves introduce confirmed admin-locale debt.

PR #83 remains open/not mergeable on head b841cd537aa2f802360b2b63e0a879cab81183cd. Against synchronized main it is now 1 commit ahead / 120 commits behind, merge base dd325cb1629f77fdac7f6494be8b752322812024, and diff scope is still exactly two files: .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po. No requested reviewers, submitted reviews, or inline review threads are present. Old-head workflows remain unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor and Auto Format pass; Preview Releases is the historical failure; Dependabot Auto-Approve is skipped.

I did not force-rewrite the translation branch in this run. A safe refresh now requires deterministic msgid-based reconstruction from the new generated 44114afd… catalog while preserving reviewed Indonesian msgstr, translating every newly empty/fuzzy entry, and then running Lingui extraction/compile plus zero-empty/fuzzy, placeholder/ICU/tag parity, diff-scope, lint, typecheck, tests, build and refreshed-head Actions. The connected GitHub API can move refs and replace files, but this generated catalog is too large to reconstruct/validate safely without an executable repository workspace; doing a partial contents-API edit would risk dropping generated references or reviewed translations.

No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): material merge-horizon change.

emdash-cms/emdash#2919 (feat(core): lock a content entry while someone is editing it) has been reconciled with current upstream main and is now mergeable again. Its head is 52d7274590c1e803fccfb6c95ee886fa5b8f4e94, based on current main/locale extraction 44114afd391ea0738bf95b4688d59513d2cb6347.

Root cause of the previous dirty/not-mergeable state was an actual conflict in packages/admin/src/components/ContentEditor.tsx: upstream emdash-cms#2972 changed the onPublishedAtChange prop while emdash-cms#2919 had moved ContentSettingsPanel inside its read-only fieldset/scroll structure. The branch author merged current upstream, kept the lock/read-only structure, and re-applied the upstream prop change. The refreshed head is now fully green across CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, and Auto Format; Dependabot Auto-Approve is skipped.

This matters to the Indonesian refresh because emdash-cms#2919 carries multiple Lingui-wrapped admin messages for edit locking/read-only/take-over/error states. If it merges before #83 is rebuilt, the following post-merge extraction will create another real Indonesian locale delta. Do not add those strings speculatively before they land; reconcile against the authoritative extraction produced after merge.

Current emdash-cms/emdash:main and ahliweb/emdash-awcms:main remain synchronized at 44114afd391ea0738bf95b4688d59513d2cb6347; no new locale extraction has landed yet. PR #83 itself remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no submitted reviews, requested reviewers, or unresolved review threads. CLA remains contributor-owned and was not signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): upstream and fork main remain synchronized at 44114afd391ea0738bf95b4688d59513d2cb6347; no newer commit or locale extraction has landed.

New CI-readiness risk: upstream issue emdash-cms#3015 reports a time/width-dependent E2E failure in content-actions.spec.ts:349. The assertion counts Range.getClientRects() for a truncate/white-space: nowrap publication-date label; when the sibling formatted date is wider (for example Sep 10, 2026, 12:54 AM), Chromium can fragment the ellipsized text run into two rects even though it has not wrapped to two visual lines. The same unchanged admin code passes when the formatted time is narrower. This caused failures in both Node and Cloudflare E2E shards on a documentation-only PR, including retries.

Root cause is therefore a brittle layout assertion coupled to clock-dependent formatted date width, not translation behavior. If refreshed #83 CI hits this failure before emdash-cms#3015 is fixed upstream, it should be treated as an upstream E2E flake rather than evidence of an Indonesian catalog regression, while still requiring a green rerun/fix before merge-readiness is claimed.

PR #83 itself remains open/not mergeable at b841cd537aa2f802360b2b63e0a879cab81183cd, 1 commit ahead / 120 behind current main, with the same 2-file translation scope and no reviews/review requests/unresolved threads. No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): i18n-adjacent CI change after the previous check.

Upstream PR emdash-cms#2903 (fix(admin): keep locale menus consistent with the admin theme) advanced to 6ff11955ed28e0d5f5a4df6c6a4fcaeddbc3d5fa with a test-only commit, test: isolate locale filter E2E data. The PR had already merged current upstream main / canonical extraction 44114afd391ea0738bf95b4688d59513d2cb6347; the new commit isolates empty/populated locale-filter cases into per-test collections and adds cleanup/recovery for posts and collections after failures.

Root cause: the locale-filter E2E coverage was sharing mutable collection/post state strongly enough that failures or partially completed creates could contaminate later assertions. The fix changes test isolation/cleanup, not locale behavior or user-visible messages. It therefore adds no Indonesian msgid debt and does not require a PO rewrite for #83.

The refreshed emdash-cms#2903 head is currently mergeable and all current-head workflows are green: CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, and Auto Format; Dependabot Auto-Approve is skipped. This reduces the chance that future #83 refreshed-head failures in locale-filter E2E are caused by upstream fixture contamination, but #83 still needs deterministic reconciliation against 44114afd… (or a later authoritative extraction if another i18n-affecting PR lands first).

Upstream and fork main remain synchronized at 44114afd391ea0738bf95b4688d59513d2cb6347; PR #83 itself remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no requested reviewers, submitted reviews, or review threads. No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): locale-related upstream PR emdash-cms#2903 (fix(admin): keep locale menus consistent with the admin theme) remains open and mergeable at 6ff11955ed28e0d5f5a4df6c6a4fcaeddbc3d5fa, with all current-head workflows green (CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, Auto Format; Dependabot Auto-Approve skipped) and all existing review threads resolved.

New merge-readiness change: after maintainer ascorbic had approved the earlier revision, the author explicitly re-requested ascorbic's review at 2026-09-10T04:43:39Z following the latest test-isolation commit. The root cause is review invalidation/rereview discipline after substantive fixture/test changes, not a new code or locale failure. The current PR still reports ascorbic as a requested reviewer, so emdash-cms#2903 is awaiting maintainer re-review even though GitHub reports it mergeable and CI is green.

emdash-cms#2903 still preserves existing Lingui message IDs and has no messages.po changes, so it does not create new Indonesian translation debt by itself. Upstream and fork main remain synchronized at canonical locale extraction 44114afd391ea0738bf95b4688d59513d2cb6347; no Indonesian branch rewrite is warranted from this development alone.

CLA remains contributor-owned; no legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-10): emdash-cms/emdash:main and fork ahliweb/emdash-awcms:main remain synchronized at canonical locale extraction 44114afd391ea0738bf95b4688d59513d2cb6347; no new main commit or locale extraction has landed.

New merge-horizon development: upstream PR emdash-cms#2934 (feat(editor): make Portable Text tables complete and responsive) refreshed its branch by merging current main at head bb66635be0950e5b411d8972a351e9cad8170d62. The PR is still reported mergeable and contains wrapped admin-facing table UI strings without committed messages.po changes, so if it merges before this Indonesian refresh is rebuilt, the next canonical extraction may add or move Indonesian catalog entries.

Current-head validation on emdash-cms#2934 is not fully green yet: Format, Query Counts, Preview Releases, CodeQL, zizmor, and Auto Format pass; CI is still queued; Visual Regression failed specifically at the final Fail on drift step after pnpm locale:extract, the build, and the visual-regression measurement itself all completed successfully. This indicates detected baseline drift/artifact difference on the refreshed branch rather than an extraction/build failure. The exact visual delta still needs upstream triage before emdash-cms#2934 is merge-ready.

No translation-branch rewrite is appropriate from this event alone because upstream main and the authoritative extraction have not changed. PR #83 remains on stale head b841cd537aa2f802360b2b63e0a879cab81183cd, not mergeable, with no submitted reviews or unresolved review threads. CLA remains contributor-owned; no legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-10): material CI-readiness follow-up on emdash-cms#2934.

emdash-cms#2934 advanced from bb66635be0950e5b411d8972a351e9cad8170d62 to 03cd2a09fb1bd193609d2efed70775cfcd654de6 with test(e2e): avoid range fragment layout flake. This directly addresses the failure mode tracked in upstream issue emdash-cms#3015: content-actions.spec.ts previously counted Range.getClientRects() on a label that is explicitly white-space: nowrap, so Chromium ellipsis fragmentation could be misclassified as a two-line wrap depending on formatted date width. The patch now treats a computed nowrap label as one line and falls back to rect counting only when wrapping is actually allowed. That is a narrowly scoped test correction, not a locale/runtime behavior change.

The same commit also tightens the table-resize test assertion without changing production behavior. Current emdash-cms#2934 head remains mergeable. Format, Query Counts, Preview Releases, CodeQL, zizmor, and Auto Format pass; CI is still queued. Visual Regression still fails, but the bot now identifies four intended/changed admin screens (content-editor-ltr, content-editor-rtl, content-new-ltr, content-new-rtl) and explicitly requires maintainer /accept-baselines if those Linux snapshot deltas are intended. Therefore the remaining visual failure is a baseline-acceptance/review decision, not the earlier emdash-cms#3015 range-fragment flake.

This reduces one known CI-flake risk for a future refreshed Indonesian PR, but it does not change the authoritative locale extraction: upstream and fork main remain synchronized at 44114afd391ea0738bf95b4688d59513d2cb6347. PR #83 itself remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no submitted reviews or review threads and its two-file translation scope unchanged. No Indonesian branch rewrite is warranted from this development alone.

CLA remains contributor-owned; no CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): material CI-readiness follow-up on emdash-cms#2934.

emdash-cms#2934 advanced again after the previous check, from 03cd2a09fb1bd193609d2efed70775cfcd654de6 through two test-only commits to fdb792d08abf5a98bcf1fe3a18857612cede5b72:

• 5df463b203b85aee7ca04cd680d8af879e917c04 (test(e2e): scope table insertion announcement)
• fdb792d08abf5a98bcf1fe3a18857612cede5b72 (test(e2e): stabilize table deletion announcements)

Both changes are narrowly scoped to e2e/tests/portable-text-table.spec.ts. They replace assertions against page.getByRole("status").first() with exact-text attachment checks for Table inserted, Row deleted, and Column deleted. Root cause: the page can contain multiple live/status regions or announcements, so selecting the first status element is order-sensitive and can bind to the wrong announcement even when the intended operation succeeded. The production table behavior and localization strings are unchanged; this is test-selector stabilization, not a runtime or Indonesian catalog change.

Current emdash-cms#2934 remains open and mergeable. On the newest head, Format, Preview Releases, CodeQL, zizmor, and Auto Format are already green; Query Counts and Visual Regression are still in progress and CI is queued at the time of this check. Therefore the branch is not yet fully revalidated on the latest head.

Upstream emdash-cms/emdash:main and fork ahliweb/emdash-awcms:main remain synchronized at canonical locale extraction 44114afd391ea0738bf95b4688d59513d2cb6347; no new main commit or locale extraction landed in this interval. PR #83 itself remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no submitted reviews or unresolved review threads and the same two-file translation scope. No Indonesian branch rewrite is warranted from these test-only commits alone.

CLA remains contributor-owned; no CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): material upstream movement and one safe maintenance action completed.

• emdash-cms/emdash:main advanced from 44114afd391ea0738bf95b4688d59513d2cb6347 to 2bc505eaa325848e6d15a08380e4ca7ad21ad6d4 via merged i18n(pt-BR): complete Brazilian Portuguese admin catalog emdash-cms/emdash#2971 (i18n(pt-BR): complete Brazilian Portuguese admin catalog) plus the bot extraction commit 2bc505e... (chore: extract locale catalogs [skip ci]).
• I fast-forwarded ahliweb/emdash-awcms:main to 2bc505eaa325848e6d15a08380e4ca7ad21ad6d4 without force. The fork is synchronized with upstream again.
• Comparing 44114afd... to 2bc505e... shows only .changeset/complete-pt-br-admin-translation.md and packages/admin/src/locales/pt-BR/messages.po changed. Therefore this new extraction introduces no Indonesian catalog delta; it is locale-specific follow-up to the pt-BR translation merge, not new id translation debt.
• Against synchronized main, i18n/id-refresh-2026-08-26 remains diverged at 1 commit ahead / 122 commits behind, merge base dd325cb1629f77fdac7f6494be8b752322812024; diff scope is still exactly .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po. PR i18n(id): refresh Indonesian admin translations #83 remains open/not mergeable, with no requested reviewer, submitted review, or review thread. Historical head checks are unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor, and Auto Format pass; Preview Releases remains the old failure; Dependabot Auto-Approve is skipped.

A second merge-horizon development is upstream emdash-cms#1944 (fix(i18n): localize invite, magic-link, and recovery emails), which has been actively refreshed today and is now mergeable on d9488f623683a126c3b497e86c249bfa0d753e0b. It introduces module-scope Lingui descriptors for invite/magic-link/recovery email copy, so if it merges before #83 is rebuilt, the subsequent canonical extraction will add real Indonesian email-message requirements.

The important technical follow-up in emdash-cms#1944 is commit 6770bc5aaf68caa4cc6c1e0e18a9b9e0c28e3a19: the previous server graph imported the admin locales barrel, which pulled React/Kumo and Lingui macro/runtime code into the Cloudflare workerd path and caused packed-install cold start failure (exports is not defined). The fix creates server-safe ./locales/config and ./locales/emails build/subpath exports and pre-bundles @lingui/core. Commit d9488f6... then clarifies the locale-helper boundary after bot review. Current emdash-cms#1944 checks are partially complete: Format, Query Counts, Preview Releases, CodeQL and zizmor are green; CI is pending and Visual Regression/Auto Format are still running. Bot review says the architecture is sound and previous RTL/BCP-47 concerns are addressed, although some review threads remain technically unresolved in GitHub.

No Indonesian PO rewrite was attempted in this run. The new main extraction does not touch id, while #83 already has unresolved translation debt from earlier authoritative extractions; a safe refresh still requires deterministic whole-catalog exact-msgid reconciliation and full refreshed-head validation rather than a partial contents edit.

No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-10): emdash-cms/emdash:main advanced from 2bc505eaa325848e6d15a08380e4ca7ad21ad6d4 to d9d30ebebb61355f516878e135444fd33ecb4aeb (6 commits). I fast-forwarded ahliweb/emdash-awcms:main to the same SHA without force, so the fork is synchronized again.

The meaningful i18n change is that emdash-cms#2919 (feat(core): lock a content entry while someone is editing it) has merged as b1ccecd5..., followed by locale extraction 23a6ba27.... The extracted Indonesian catalog now contains newly generated entry-lock msgids with empty msgstr values (confirmed for Another editor; the source also adds This entry is open somewhere else, Open read-only, Take over, Taking over..., Read-only, You no longer hold this entry, and the related explanatory Trans messages). This turns the previously anticipated locale debt into an actual reconciliation requirement for #83.

Root cause: emdash-cms#2919 introduces a per-entry lease with takeover/read-only UX so concurrent editors are warned before conflicting writes. The extraction correctly materializes those new Lingui messages; Indonesian has not yet been translated for them because #83 predates the merge/extraction.

Branch state after fork sync: i18n/id-refresh-2026-08-26 remains 1 commit ahead / 128 commits behind main, merge base dd325cb1629f77fdac7f6494be8b752322812024, and scope remains only .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po. PR #83 is still not mergeable and has no submitted review, requested reviewer, or unresolved review thread. Old-head checks remain green except the historical Preview Releases failure; refreshed-head CI is still required after reconciliation.

Two new upstream reports also appeared today but are outside this translation PR's scope: emdash-cms#3022 reports the admin CSP blocking eval/blob workers required by current UI behavior, and emdash-cms#3023 reports missing Tailwind utilities in the shipped admin stylesheet causing dialogs to render off-screen. They may affect upstream runtime/package readiness but do not currently add Indonesian msgids.

I did not rewrite the translation branch in this run because a safe deterministic rebuild must preserve all previously reviewed translations while rebasing onto the new canonical extracted catalog; the connected GitHub write surface available here does not provide a conflict-aware cherry-pick/rebase operation for the existing large PO file. No partial PO edit was made. CLA remains contributor-owned and was not signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced three documentation-only commits from 573230f539e03ea99e6c22f2cd7f704a4abd25d5 to f35c7d491c556f3a46f06521221187dd23a1d18b (emdash-cms#3035, emdash-cms#3030, emdash-cms#3029). I fast-forwarded ahliweb/emdash-awcms:main to the same SHA without force.

No newer locale extraction landed: 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest chore: extract locale catalogs [skip ci] commit. The three new main commits are documentation/guidance changes, so they do not add Indonesian msgid debt or justify a partial PO rewrite.

After synchronization, i18n/id-refresh-2026-08-26 remains translation-only in scope (changeset + packages/admin/src/locales/id/messages.po) and is now 1 commit ahead / 135 commits behind current main, with merge base dd325cb1629f77fdac7f6494be8b752322812024. PR #83 remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no requested reviewers, submitted reviews, or inline review threads. Old-head Actions are unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor and Auto Format pass; Preview Releases is the historical failure; Dependabot Auto-Approve is skipped. Fresh-head validation is still required after deterministic catalog reconciliation.

New upstream security/privacy item emdash-cms#3049 reports a delete-integrity flaw: the media delete route removes the DB row first, ignores a failed storage-delete boolean, and the public media-file route serves directly from storage without requiring a DB row. Root cause: non-atomic DB/storage deletion plus orphan objects remaining publicly addressable. This is not caused by the Indonesian catalog and adds no confirmed locale delta yet, but it is a meaningful upstream security/privacy issue to monitor because any admin-facing remediation could later add localized error/status messages.

Related new security hardening PRs emdash-cms#3050 (SSRF validation before plugin HTTP dispatch), emdash-cms#3051 (bind Workerd bridge access to active plugin credentials), and emdash-cms#3052 (keep content attribution under host control) currently state that they change no admin UI strings; they therefore do not add confirmed Indonesian catalog debt at this point.

No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced three additional documentation-only commits from f35c7d491c556f3a46f06521221187dd23a1d18b to 2579547a55f92692d49b1ee005dbc6641b6d17f9 (emdash-cms#3042, emdash-cms#3044, emdash-cms#3038). I fast-forwarded ahliweb/emdash-awcms:main to the same SHA without force and verified the fork now matches upstream.

No newer locale extraction landed: 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest commit touching packages/admin/src/locales/id/messages.po. The three new main commits are documentation-only, so they add no Indonesian msgid debt and do not justify a partial PO rewrite. After sync, i18n/id-refresh-2026-08-26 is 1 commit ahead / 138 commits behind current main, merge base dd325cb1629f77fdac7f6494be8b752322812024; PR scope remains the translation changeset plus packages/admin/src/locales/id/messages.po.

New bot-generated upstream work is relevant to the merge horizon:

• Draft PR Add build-time admin.locales allowlist emdash-cms/emdash#3056 implements issue Ship only the admin locales a site uses (build-time locale allowlist) emdash-cms/emdash#2973 with a build-time admin.locales allowlist. Root cause: all admin locale catalogs are currently retained in the deployed bundle even when a site uses only a subset; the reported production measurement was about 3.01 MB / 29% of the Worker bundle. The candidate filters the runtime locale list and redirects excluded source/dist locale imports to English so fallback still works. It is open/mergeable as a draft, preview build/deployment succeeded, while the normal CI/Format/Query Counts/Visual Regression/CodeQL/zizmor/Preview Releases checks were still queued at this check. It changes locale-loading infrastructure but introduces no confirmed new Indonesian UI msgid in its current patch.
• Draft PR fix(core): separate revision author from entry owner and pass actor to content hooks (#2881) emdash-cms/emdash#3057 implements issue Nothing records who changed content: admin revisions have a null author, the MCP attributes by reassigning entry ownership, and hooks get no actor emdash-cms/emdash#2881. Root cause: revision attribution, entry ownership, and acting-user identity were conflated: admin revisions could get NULL authors while MCP updates could silently reassign ownership. The bot patch passes actor metadata separately, keeps owner changes explicit, and exposes actor id/role to content hooks. It is open/mergeable as a draft; normal validation workflows were still queued. It states no admin UI change, so there is no confirmed Indonesian locale delta from it.

PR #83 itself remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no requested reviewers, submitted reviews, or inline review threads. Old-head Actions remain unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor and Auto Format pass; Preview Releases is the historical failure and Dependabot Auto-Approve is skipped. Fresh-head validation is still required after deterministic catalog reconciliation.

No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): upstream main advanced one additional documentation/reference-inventory commit to efb7677f2ad7d87f53e52cf73be66bb203b15857 (emdash-cms#3048). I fast-forwarded ahliweb/emdash-awcms:main to the same SHA without force and verified the fork now matches upstream. The commit is docs/tooling-only and does not touch admin locale catalogs, so 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest canonical Indonesian locale extraction and no PO rewrite is warranted from this commit alone.

Merge-horizon update for draft upstream PR emdash-cms#3056 (admin.locales allowlist): its first candidate had a real Format CI failure. Root cause was limited to .changeset/admin-locale-allowlist.md: a JavaScript code block used spaces where the repository formatter requires tabs. emdashbot revised the candidate without source-code changes; local verification reports pnpm format:check, pnpm lint:quick, workspace typecheck, 335 Astro integration tests, and 28 focused locale tests passing. A fresh Cloudflare preview build/deployment for head 4b123fb65fefb40748b9ce6f401dddf7f2fbfb09 succeeded. The normal GitHub Actions rerun is still queued/pending, so emdash-cms#3056 is not yet fully validated and remains a draft. This changes locale-loading infrastructure but still introduces no confirmed new Indonesian msgid.

PR #83 itself remains unchanged on b841cd537aa2f802360b2b63e0a879cab81183cd; no requested reviewers, submitted reviews, or inline review threads are present. Deterministic catalog reconciliation and fresh-head CI are still required before merge readiness. No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): synchronized fork main by fast-forward from efb7677f2ad7d87f53e52cf73be66bb203b15857 to upstream 006d50c4afb6d82cd19a6e3ca850ffc3dc759019. The two new upstream commits are documentation/reference-inventory work only and do not touch packages/admin/src/locales/id/messages.po; 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest locale extraction.

After sync, i18n/id-refresh-2026-08-26 remains translation-only in diff scope (changeset + Indonesian catalog), now 1 commit ahead / 141 commits behind current main, merge base dd325cb1629f77fdac7f6494be8b752322812024. PR #83 remains open and not mergeable, with no maintainer reviews, review requests, or unresolved review threads. Historical head checks are unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor and Auto Format passed; Preview Releases remains the historical failed workflow. Fresh-head validation is still required after deterministic reconciliation.

Merge-horizon update: upstream emdash-cms#3046 advanced from aa54f03e... to efdd6bf7857e2fd9aae47bfd0afc9c7bfe2b5bcf. The follow-up only relocates applyTermFilters so applyBylineFilter keeps its own JSDoc and tidies comments/changeset punctuation; it does not add or alter localized UI strings beyond the two already identified in emdash-cms#3046. All fresh-head workflows currently report action_required with zero jobs executed, so this is an approval/execution gate rather than evidence of test/lint/build failure. emdash-cms#3046 remains open and mergeable and would still create future Indonesian msgids if merged and extracted before this PR is rebuilt.

Current canonical Indonesian catalog on main visibly contains untranslated entries introduced after this branch (for example Media Library delete/upload strings), confirming that deterministic catalog reconciliation remains necessary; no partial PO edit was applied because it would leave the branch structurally stale.

No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced by one docs-only commit to a9c01a956bca80786419b8264712145c97f7df42 (docs: improve content authoring journey (#3040)). The fork main has been fast-forwarded non-force to the same SHA. The Indonesian locale extraction has not changed: 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest commit touching packages/admin/src/locales/id/messages.po. The translation branch is now 1 commit ahead / 142 commits behind current main; its diff scope remains exactly .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po.

New merge-horizon development: upstream emdash-cms#2156 (feat(core): built-in SMTP email transport for generic SMTP credentials) advanced to 8543c8253ad13d6bb0f89e7378d29dacd1c5f225 and is currently open/non-draft/mergeable. The author reports the previous Node-adapter CI failures were caused by a literal import("cloudflare:sockets"), which Rolldown could not resolve during Node astro build. The fix routes sockets through a generated virtual:emdash/sockets adapter module, exporting Cloudflare sockets only under the Cloudflare adapter and falling back to Node networking elsewhere. The same update also wraps the remaining SMTP/Cloudflare admin-form placeholders for translation. Therefore, if emdash-cms#2156 merges before this PR is reconciled, the next canonical extraction is expected to add/adjust Indonesian msgids. Fresh GitHub Actions for emdash-cms#2156's new head are still queued, so it is not yet fully validated.

PR #83 itself has no new maintainer comments, review requests, submitted reviews, or unresolved review threads. Its historical head checks remain unchanged; refreshed-head validation is still required after deterministic PO reconciliation. No CLA or other legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced one documentation-only commit from a9c01a956bca80786419b8264712145c97f7df42 to 2d3f8960cc4f857fe53dd433f5ffb8241442969f (docs: correct plugin installation and registry guidance (#3036)). I fast-forwarded fork main non-force to the same SHA and verified synchronization. Because the new commit is documentation-only, it does not add an admin locale extraction or Indonesian msgid debt; the translation branch itself remains unchanged and should still be reconciled deterministically rather than patched partially.

Merge-horizon update: upstream emdash-cms#2156 (feat(core): built-in SMTP email transport) now has all current GitHub Actions green (CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, Auto Format). GitHub currently reports it mergeable but mergeable_state: unstable and rebaseable: false, with review/needs-rereview; its head is still based on an older main. This matters to #83 because emdash-cms#2156 contains Lingui-wrapped admin SMTP/Cloudflare Email strings, so if it is refreshed and merged before this PR, the next canonical extraction may add Indonesian msgids.

A newly active bot triage on upstream issue emdash-cms#3023 is investigating missing Tailwind utilities in the shipped admin stylesheet (top-8, sm:top-16, max-w-[calc(100vw-2rem)]) that can render dialogs off-screen. The reported root cause is package CSS generation/content scanning rather than localization. The bot has identified the cause but has not yet posted its final triage disposition, so there is no translation-branch action to take from it yet.

PR #83 still has no submitted maintainer review, review request, or unresolved inline review thread. CLA remains contributor-owned; no legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced by two documentation-only commits from 2d3f8960cc4f857fe53dd433f5ffb8241442969f to b989bb9de2ac9ce5c792cbab2873b1c0689232cf (docs: rebuild the EmDash front door (#3045) plus its direct predecessor). I verified the old fork head is the merge base of the new upstream head, then fast-forwarded ahliweb/emdash-awcms:main non-force to b989bb9de2ac9ce5c792cbab2873b1c0689232cf. The compare is docs-only; no admin locale files changed.

No newer Indonesian locale extraction landed: 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest commit touching packages/admin/src/locales/id/messages.po. Therefore no partial messages.po rewrite was applied.

After synchronization, i18n/id-refresh-2026-08-26 is 1 commit ahead / 145 commits behind main, with merge base dd325cb1629f77fdac7f6494be8b752322812024; diff scope is still exactly .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po. PR #83 remains open/not mergeable with no requested reviewers, submitted reviews, or inline review threads. Historical head Actions remain unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor, and Auto Format pass; Preview Releases is the historical failure; Dependabot Auto-Approve is skipped.

Merge-horizon update: upstream emdash-cms#3046 (feat: filter the content list by taxonomy term) advanced to e13a7f6257346341eb7b76af5388be100e26c078. GitHub now reports the PR as technically mergeable but mergeable_state: blocked and rebaseable: false, with review/needs-review, overlap, and XL-scope labels. Its two admin strings remain Lingui-wrapped and would create future Indonesian msgids if merged and extracted before this PR is rebuilt. The reported broad local test failures (36 files / 458 tests) reproduce identically on clean upstream main and are attributed by the author to a local better-sqlite3 native-module baseline, while the seven added taxonomy-filter tests pass; this is therefore not evidence of a regression specific to the candidate. No safe action on #83 follows from emdash-cms#3046 at this point.

No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced from b989bb9de2ac9ce5c792cbab2873b1c0689232cf to d30207d391bfb10f8a0566992e02df246dfa8c02 via emdash-cms#3060 (fix(admin): hide dashboard quick actions for hidden collections). The fork main has been fast-forwarded non-force to the same SHA.

Root cause of emdash-cms#3060: dashboard Quick Actions enumerated all manifest collections while the sidebar already filtered hidden collections, so hidden: true collections could still expose a + New … action on the dashboard. The merged fix reuses the visible-collection filtering semantics and adds regression coverage. The merged diff contains no locale catalog changes and does not introduce new translation msgids.

Canonical Indonesian extraction is unchanged: 21ec8bb32a3b9f33807897a2a6627f8672bb4815 remains the latest commit touching packages/admin/src/locales/id/messages.po.

After sync, i18n/id-refresh-2026-08-26 remains 1 commit ahead / 146 commits behind main, merge base dd325cb1629f77fdac7f6494be8b752322812024, with scope still limited to .changeset/bright-id-refresh.md and packages/admin/src/locales/id/messages.po. PR #83 remains non-mergeable. There are still no submitted reviews, requested reviewers, or unresolved review threads. Historical head checks are unchanged: all primary checks pass except the old Preview Releases failure; a fresh full validation remains required after deterministic PO reconciliation/rebase.

No partial catalog edit was applied because emdash-cms#3060 creates no new msgid and the branch still requires full deterministic reconciliation against the canonical catalog. CLA remains contributor-owned; no legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced from 0ae2f265... to 92690af1... via merged emdash-cms#2934 (Portable Text tables) followed by a fresh locale extraction. I fast-forwarded ahliweb/emdash-awcms:main non-force to 92690af1...; fork and upstream main are now synchronized.

This extraction is materially different from the prior source-reference-only churn: emdash-cms#2934 introduces many translatable table-editor messages and the current Indonesian catalog contains new empty msgstr entries, including Add row above, Add row below, table size/selection announcements, row/column deletion messages, cell merge/split feedback, and column-width actions. The extraction commit modifies the Indonesian catalog by 406 additions / 139 deletions, so PR #83 now requires a full deterministic reconciliation against 92690af1..., not a partial source-reference refresh.

Current branch comparison after the fork sync: i18n/id-refresh-2026-08-26 is 1 commit ahead / 150 commits behind main, merge base dd325cb1...; diff scope remains the translation changeset plus packages/admin/src/locales/id/messages.po. GitHub still reports PR #83 not mergeable. No submitted reviews, requested reviewers, or unresolved review threads are present. Historical CI state is unchanged (all principal checks green except the old Preview Releases failure).

Merge horizon: emdash-cms#3062 remains open/mergeable and contains additional localized admin navigation strings. Its aggregate CI currently fails only on baseline/flaky failures documented by the author: the known publication-date line-wrap E2E failure reproduces on main, and the Smoke Test failure is a cold-start timeout/retry race that caused two seed requests to overlap and hit a taxonomy UNIQUE constraint; the branch's local smoke suite passed 195/195. If emdash-cms#3062 lands before this PR is reconciled, another extraction delta is expected.

I did not force-update or partially rewrite the translation branch because that would risk dropping reviewed translations or producing a structurally inconsistent generated PO catalog. CLA remains contributor-owned; no legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-11): emdash-cms/emdash:main advanced one commit from 92690af1a6955eb06751b512ec3e66b0bffdb675 to 36a021c1185073e77da891d54a406ea9ce810826 via merged emdash-cms#3057 (fix(core): separate revision author from entry owner and pass actor to content hooks (#2881)). I fast-forwarded ahliweb/emdash-awcms:main non-force to the same SHA and verified synchronization.

Root cause of emdash-cms#3057: revision authorship, entry ownership, and the authenticated acting user were conflated. Admin saves could leave revisions.author_id null, while MCP writes could populate revision attribution by also silently reassigning ec_* .author_id. The merged fix carries actor identity separately, preserves owner semantics, exposes actor id/role to content hooks, and adds regression coverage. It changes core/plugin behavior and documentation, not admin locale catalogs or user-visible admin strings, so the canonical Indonesian extraction remains 92690af1... and no additional PO delta is introduced by this merge. The PR head's CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, and Auto Format checks all passed; Dependabot Auto-Approve was skipped.

Merge-horizon note: emdash-cms#3056 (admin.locales allowlist) has refreshed to head fc2fec1193b827782b4c33fb4f55b05ad098fd93, is open/non-draft/mergeable, and its latest Cloudflare preview build/deployment succeeded. Format, Query Counts, Visual Regression, Preview Releases, CodeQL, and zizmor are green; aggregate CI is currently queued. It still changes locale-loading infrastructure rather than adding confirmed Indonesian msgids.

PR #83 itself remains open/not mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no requested reviewers, submitted reviews, or unresolved review threads. Historical head checks remain unchanged: all principal checks pass except the old Preview Releases failure. The material translation blocker remains deterministic reconciliation against the 92690af1... extracted catalog, which contains the new Portable Text table msgids; no partial PO rewrite was applied. No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-11): upstream main advanced from 36a021c… to 4c8913057cdab82c7af66a126722525ee74ba4cb via three merged changes: emdash-cms#3065 (surface Cloudflare binding/configuration errors), emdash-cms#2351 (make Cloudflare sandboxed plugins a paid-plan Worker Loader opt-in), and emdash-cms#3050 (validate plugin HTTP destinations against SSRF rules before initial dispatch and redirects). Fork main has been fast-forwarded non-force to the same SHA.

None of these three merges changes packages/admin/src/locales/* or adds admin UI msgids, so 92690af1… remains the latest canonical Indonesian locale extraction. emdash-cms#3050 is nevertheless security-significant: allowedHosts previously checked host membership without shared SSRF validation, including redirect targets. The merged fix validates scheme, hostname, DNS-resolved addresses, and additional non-public IPv4/IPv6 ranges; its current head passed CI, Format, Query Counts, Visual Regression, Preview Releases, CodeQL, zizmor, and Auto Format.

A separate i18n merge-horizon item is now material: upstream emdash-cms#1944 (fix(i18n): localize invite, magic-link, and recovery emails) is open with review/approved but currently labeled needs-rebase and overlap. It routes system-email copy through Lingui descriptors and localizes the invite-name placeholder, so if it rebases and merges before this PR is rebuilt, the next extraction will add real Indonesian msgids. Its current head CI suite is green, but GitHub currently reports it non-mergeable due to branch freshness/overlap rather than test failure.

Current branch comparison after sync: i18n/id-refresh-2026-08-26 remains 1 commit ahead / 154 commits behind main, with merge base dd325cb…, and scope remains only .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po. No maintainer review, review request, or unresolved review thread is present on #83. Deterministic full PO reconciliation and fresh-head CI remain required; no partial PO rewrite was applied in this check because the current catalog already contains semantic translation debt and a partial generated-file edit would not be safely reproducible.

CLA remains contributor-owned; no legal agreement was signed or accepted by automation.

[ahliweb]

Upstream watch update (2026-09-12): upstream and fork main remain synchronized at b750ddf094686b82fa28557b47281251c4b8a9c3; no new locale extraction has landed.

Meaningful merge-horizon change: upstream emdash-cms#3072 (Fix the admin theme toggle) advanced to 7838e580d4320243021f8670fa45157f5cb3823f after maintainer-requested visual baseline acceptance. Its complete current workflow set is now green: CI, Visual Regression, Format, Query Counts, Preview Releases, CodeQL, and zizmor all pass (Auto Format and Dependabot are skipped). The PR remains open and mergeable. Because emdash-cms#3072 introduces the translatable admin labels Switch to dark and Switch to light, a merge would trigger a new locale extraction and additional Indonesian translation debt.

Upstream emdash-cms#1944 (localized invite/magic-link/recovery emails) remains open, mergeable, and poised to add a larger locale delta if merged. New emdash-cms#3073 adds raw plugin HTTP request/response support but explicitly changes no admin UI strings, so it does not add Indonesian translation debt.

PR #83 itself remains open/not-mergeable on b841cd537aa2f802360b2b63e0a879cab81183cd, with no requested reviews, submitted reviews, or unresolved review threads. No branch rewrite or partial PO edit was performed: current main has not moved, while two merge-ready i18n-affecting PRs remain near the merge horizon, so rebuilding now would likely become stale immediately. CLA remains contributor-owned; no legal agreement was accepted or signed.

[ahliweb]

Upstream watch update (2026-09-12): emdash-cms/emdash:main advanced from b750ddf094686b82fa28557b47281251c4b8a9c3 to 633257001b940e267ec818e5c9040f250f45fca7 via merged emdash-cms#3072 (Fix the admin theme toggle) followed immediately by chore: extract locale catalogs [skip ci]. I verified the old fork head was the merge base and fast-forwarded ahliweb/emdash-awcms:main non-force to 633257001b940e267ec818e5c9040f250f45fca7.

This extraction adds a real Indonesian translation delta: emdash-cms#3072 replaces the previous three-state theme-cycle UI with a direct resolved light/dark action and introduces Lingui messages Switch to dark and Switch to light. Both are currently present in the canonical Indonesian catalog with empty msgstr. The extraction changes each locale catalog by 8 additions / 29 deletions, reflecting those new labels plus removal/source-reference cleanup of the former theme-toggle messages. Root cause: the old system→light→dark cycle could require a click that produced no visible scheme change; the new control always toggles the visible scheme and falls back to system when the selected scheme matches OS preference.

After synchronization, i18n/id-refresh-2026-08-26 remains translation-only in scope (changeset + Indonesian catalog) and is now 1 commit ahead / 158 commits behind current main, merge base dd325cb1629f77fdac7f6494be8b752322812024. PR #83 remains open/not mergeable with no requested reviewers, submitted reviews, or inline review threads. Historical head Actions are unchanged: CI, Format, Query Counts, Visual Regression, CodeQL, zizmor, and Auto Format pass; Preview Releases remains the historical failure and Dependabot Auto-Approve is skipped.

No partial PO rewrite was applied because the branch still needs full deterministic reconciliation against the canonical catalog, which already contains earlier untranslated table/navigation strings in addition to the two new theme labels. Upstream emdash-cms#1944 (localized invite/magic-link/recovery emails) remains open and technically mergeable in the REST view, but is now marked review/needs-rereview/overlap and is not rebaseable; if it lands later it will create another material locale delta. No CLA or other legal agreement was signed or accepted.

[ahliweb]

Upstream watch update (2026-09-12): emdash-cms/emdash:main advanced by one commit to 33cb7f08de03fb7febccc72c9eb29fcf88b9c248 via merged emdash-cms#3075 (fix(core): retain plugin bundles across updates). I verified the previous fork head 633257001... is the direct parent and fast-forwarded ahliweb/emdash-awcms:main non-force to 33cb7f08....

emdash-cms#3075 is core-only: it removes update-time deletion of previous marketplace/registry plugin bundles and adds concurrency/conformance coverage. Root cause was a delayed cleanup race: overlapping update/downgrade work could delete a bundle that had become active again. Install-failure cleanup and explicit uninstall cleanup remain unchanged. Its full current workflow set was green (CI, Visual Regression, Format, Query Counts, Preview Releases, CodeQL, zizmor, Auto Format). No admin locale files or user-facing strings changed, so this commit does not add Indonesian translation debt.

New merge-horizon item: upstream emdash-cms#3076 (fix(admin): require marketplace update re-consent) is open and currently mergeable. It changes marketplace updates from effectively pre-confirmed capability/MCP consent to an unconfirmed, exact-version-pinned preflight. The admin then presents newly requested capabilities/public routes/MCP tools before confirmation. The patch introduces a new Lingui fallback message, Plugin update requires re-consent, so if emdash-cms#3076 merges it will create at least one new Indonesian msgid at the next extraction. emdashbot also raised a review suggestion that the consent dialog should display the exact reviewed version; otherwise a background update check can make the card display a newer version while the confirmation still authorizes the pinned older version. Current emdash-cms#3076 checks are mixed: Format, Visual Regression, Query Counts, Preview Releases, CodeQL, zizmor, and Auto Format pass, but aggregate CI is red because the E2E Tests check failed; the available GitHub metadata exposes the failed check but not its underlying annotation/log detail, so I am not attributing that failure to the implementation without evidence.

After fork sync, i18n/id-refresh-2026-08-26 remains 1 commit ahead / 159 commits behind current main, merge base dd325cb1629f77fdac7f6494be8b752322812024, and its diff scope remains only .changeset/bright-id-refresh.md plus packages/admin/src/locales/id/messages.po. PR #83 remains open/not mergeable with no requested reviewers, submitted reviews, or inline review threads. Historical head workflows remain unchanged: CI, Visual Regression, Format, Query Counts, CodeQL, zizmor, and Auto Format pass; Preview Releases remains the old-head failure.

No partial PO rewrite/rebase was applied in this check. Current main did not receive a locale extraction after emdash-cms#3075, while emdash-cms#3076 is an active i18n-affecting merge candidate and the branch still requires deterministic full reconciliation against the canonical catalog. CLA remains contributor-owned; no legal agreement was signed or accepted by automation.