[LostInInaka]

The launcher ships thirty translated languages plus English, and Japanese is not one of them — which is
odd for a launcher whose tested list has a Sharp AQUOS and a Sony BRAVIA on it. This adds values-ja with
all 145 strings and the one plurals resource, written for a TV rather than for a desktop: short labels, 設定 and アプリ and
壁紙 where Japanese says them, product names left alone, and the long help strings cut to two or three
sentences so they fit on a tile.

• date_format_home is not in the file: it is translatable="false" and aliases @string/date_format,
  which is translated.
• library_app_count keeps quantity="other" only, as values-ko does — Japanese has no plural.
• Every format argument was checked against the source (%1$s, %1$d, %2$s, %1$.0f, %1$d%%): same
  set, same order. race_next keeps %1$s before %2$s so it reads 「予選まで3時間」.
• Wired the way the tests ask: res/xml/locales_config.xml, the Languages list in docs/launcher.md,
  NAMES in tests/test_locales_registered.py, and the generated counts. tools/gen_tables.py now reads
  416 TV models, 106 brands and 32 languages, --check exits 0, and the card and its alt text were redrawn
  to match.
• No fonts.xml: zh and ko ship without one and Android's fallback covers the glyphs.

The size gate, up front. On today's main (3cd274f8) the release APK is 587,437 B without this
locale and 595,709 B with it — 5,885 B past the 589,824 B gate in ci.yml, and main itself carries
only 2,387 B of headroom before this branch starts. docs/workflows/translation.md step 7 says a locale that does
not fit is the maintainer's call rather than something to trim strings for, so nothing has been trimmed.
This is #2545's situation rather than a new one (its 5,149 B over is a shade less than this one's), and I read your answer there — it stays open while you look
for room in the launcher's own images — so this one is here to wait too, and I'll keep it rebased until
there is room. Sizes are from a clean assembleRelease on my machine; your main is documented at
585,489 B where mine builds 587,437 B from the same tree, so treat the figures as ±2 KB and the delta
between them as the real number.

Everything else is green: 2788 tests, one failure, and that one
(test_disable_ps1_refuses_the_bare_names_before_touching_the_tv, a PowerShell colour case) fails
identically on main in the same environment — I ran both to be sure.

It has been through a revision pass on how the Japanese reads, which is where much of the diff comes from: the Dussehra
greeting was ドゥシェラ and is now ダシェラ; デバイスなし for "not connected" became 未接続; TVのシステム設定
became システム設定; 「12時間」/「24時間」 became 「12時間表示」/「24時間表示」; the home-settings help lost its
machine-translated clause order; and a line in README.ja.md that still said the launcher has no Japanese
now says how to set the app's own language to ja. 有線LAN, スタンバイ and フリー走行 were kept
deliberately, so a native speaker may still want to argue about three words.

[github-actions]

maintainer-bot review · suggests merge (p2)

Thank you for the careful Japanese translation: the labels are short enough for a TV, and every format argument was checked against the source. This looks ready to merge as a strings-only launcher translation.

Checks

• Note: possibly collides with Launcher: add Slovenian translation #2545 (same lines of README.de.md, readme_stats.py; whichever merges second needs a rebase)
• Note: possibly collides with Add NVY TV with GAIA OS #2569 (both PRs modify README.de.md line 23 (image alt text) and tools/readme_stats.py line 74 (stats tuple))
• Note: possibly collides with Add Russian Lang #3164 (edit to README.tr.md, README.uk.md, and README.md language list/link sections)
• Note: possibly collides with README: the four translated front pages state the 228 projects SHOWCASE.md holds #3577 (both PRs modify README.uk.md and other README.* files in overlapping regions (language counts and project counts))
• Note: possibly collides with themes: add Seasonal Glass day/night artwork #3585 (README.tr.md and README.uk.md both update the language count in the same table row, but PR Launcher: Japanese (values-ja), every translatable string #3040 changes it to 32 (adding Japanese) while PR themes: add Seasonal Glass day/night artwork #3585 keeps it at…)
• Note: possibly collides with launcher, themes: a terminal-hacker look, in text only #3632 (the same lines of README.de.md, README.tr.md, README.uk.md; whichever merges second needs a rebase)
• Launcher: 1 lines changed in launcher/ (tests not counted)
• OK: no new permission, no network use
• Size: size/M (+170 -17, 16 files)

This is a suggestion; a maintainer makes the final call. A maintainer can comment /bot review to run it again.

_{Reviewed by maintainer-bot · runs on 0G Private Computer}

[Gathin23]

Thank you @LostInInaka, this reads like it was written for a TV: 未接続, 12時間表示, and every format argument checked against the source. It adds about 8.3 KB and goes 5.9 KB past the size gate. That makes it the same maintainer call as #2545, so we're keeping it open alongside that one rather than trimming it. We can't promise when room will come, and thank you for offering to keep it rebased.

[LostInInaka]

@Gathin23 — Rebased onto today's main (d7b0d4a5) and re-measured, since the numbers in the body moved with it:

• release APK without this locale: 588,061 B
• with it: 596,333 B — 6,509 B past the 589,824 B gate
• main itself now carries 1,763 B of headroom

Counts regenerated from the tree: 427 TV models, 111 brands, 32 languages, card and alt text redrawn, gen_tables --check clean. Nothing trimmed, and I will keep it rebased while it waits — no promise needed in return.

[LostInInaka]

@Gathin23 — the room you were waiting for is in review: #3156 takes 18,764 B back by shipping the bundled Outfit variable font clipped to the four weights res/font/outfit*.xml actually declares, with the glyphs the app can draw. On today's main that is 588,061 → 569,297 B, 20,527 B under the gate.

So I measured this branch with it. Same head as here (aa4586d8), only the font swapped for the trimmed one:

build	APK	vs 589,824 B
this branch, as it stands	596,333 B	6,509 over
this branch + the font trim	577,569 B	12,255 under

314 launcher cases green, nothing trimmed, gate untouched — docs/workflows/translation.md step 7 stays honoured, because the bytes came out of a typeface rather than out of Japanese.

Which means the gate no longer has to decide this PR, if #3156 goes first. I will keep this rebased either way; if the order happens to be the other way round I will rebase onto it and re-measure.

One housekeeping note on the collision the bot raised: #2545 also takes the language count from 31 to 32, so whichever of the two merges second regenerates the counts (tools/gen_tables.py) — a command, not a design question.

[LostInInaka]

One correction to the figure I gave you yesterday, because it moved: the typeface trim now keeps weight 100 in its axis (the bare @font/outfit that font_body points at draws the default instance, which is 100 — clipping it away would have thickened every tile subtitle), so it is 18,964 B rather than 18,764.

Same head as here (aa4586d8), font swapped, rebuilt and measured: 596,333 → 577,369 B, 12,455 B under the 589,824 B gate, 314 launcher cases green. The comparison and the conclusion are unchanged; only the last four digits moved.

[LostInInaka]

@Gathin23 — a heads-up that costs you nothing to ignore: this branch is now 402 commits behind main (3c122bc0), so when the typeface PR lands it will need a real rebase, not a one-click update. Waiting is still the right call — the size question that blocks this resolves once #3156 is in — but I did not want it to look like it sits one button away from green. I will do the rebase myself the day that lands. I have deliberately not re-measured this branch against today's main, so there is no new figure here to chase.