docs: add Simplified Chinese and Japanese READMEs - #317
Merged
Conversation
Three of the channels we want to reach expect a README in their language before they will consider a project: HelloGitHub's monthly issue, Ruan Yifeng's weekly, and the V2EX/Juejin/Zhihu posts that follow from them. An English-only repo is filtered out early, so this is a prerequisite rather than a nicety. Deliberately not a full translation. The English README is 830 lines and changes several times a week; a complete mirror would drift, and a stale Chinese README is worse than none. These carry the core a reader needs to decide - the positioning, quick start, the ten engines, install paths, where the paid line falls, and the test story - and send the deep sections (development databases, seed connections, deployment recipes) to the English docs. Two additions rather than translations, aimed at that audience: how Redis is mapped onto a SQL-shaped interface (SCAN over key prefixes, never KEYS *), and an explicit note that issues and pull requests in Chinese or Japanese are welcome. The English README gains a language switcher, without which neither file is reachable.
House style for anything written into this repo: no emoji, no em dashes, no --- separators. The two files went in carrying 18 em dashes and 18 rules between them. Section headings already do the separating, and the dashes are replaced with the punctuation the sentence actually wanted.
Contributor
There was a problem hiding this comment.
Pull request overview
Adds concise Simplified Chinese and Japanese project introductions for regional developer communities.
Changes:
- Adds localized quick starts, features, installation, and contribution guidance.
- Documents Redis’s SQL-shaped interface.
- Adds README language navigation.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
README.md |
Adds the language switcher. |
README_zh.md |
Adds the Simplified Chinese README. |
README_ja.md |
Adds the Japanese README. |
Suppressed comments (4)
README_zh.md:111
- The AI panel is rendered even when no API key is configured, and Ollama is intentionally valid without a key. This sentence therefore gives incorrect setup and UI expectations; distinguish providers that require keys instead.
**不配置密钥，这个面板根本不存在**，默认不会有任何数据离开你的网络。
README_zh.md:141
- This promises an organization-wide query audit log, but query executions are not written to the server audit buffer:
query_executionis only an unused event type, while the server currently records maintenance events. Remove this claim or implement centralized query auditing before advertising it.
7 个标签页的监控面板（概览、性能、查询、会话、表、存储、连接池）、时序趋势图、5-60 秒可调自动刷新、阈值告警配色，以及一键 `VACUUM` / `ANALYZE` / `REINDEX` / `UPDATE STATISTICS` / `DBCC CHECKDB` / `ALTER INDEX REBUILD`。全组织的查询审计日志一并提供。
README_ja.md:111
- The AI panel is rendered even when no API key is configured, and Ollama is intentionally valid without a key. This sentence therefore gives incorrect setup and UI expectations; distinguish providers that require keys instead.
**キーを設定しなければ、このパネルは存在しません。** 既定では何もネットワークの外に出ません。
README_ja.md:141
- This promises an organization-wide query audit log, but query executions are not written to the server audit buffer:
query_executionis only an unused event type, while the server currently records maintenance events. Remove this claim or implement centralized query auditing before advertising it.
7タブのモニタリング（概要、パフォーマンス、クエリ、セッション、テーブル、ストレージ、コネクションプール）、時系列トレンドグラフ、5〜60秒で調整可能な自動更新、しきい値による色分けアラート、そしてワンクリックの `VACUUM` / `ANALYZE` / `REINDEX` / `UPDATE STATISTICS` / `DBCC CHECKDB` / `ALTER INDEX REBUILD`。組織全体のクエリ監査ログも含みます。
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
|
|
||
| ### AI 助手（可选，用你自己的模型） | ||
|
|
||
| - **不绑定厂商**：默认 Gemini 2.5 Flash，同样支持 OpenAI、Claude，或 **本地模型**（Ollama / LM Studio）。 |
|
|
||
| ### AIアシスタント（任意・自分のモデルで） | ||
|
|
||
| - **ベンダー非依存**：既定はGemini 2.5 Flash。OpenAI、Claude、**ローカルLLM**（Ollama / LM Studio）にも対応。 |
| これを本気で受け取ると、好みの問題ではなく仕様になります。 | ||
|
|
||
| - データも同僚も自分のマシン上にはいないので、エディタはブラウザで動く必要がある。 | ||
| - クエリが必要になる障害は、あなたがノートPCを開くまで待ってくれないので、スマートフォンから届く必要がある。 |
…lized READMEs Review of #317 found that the abridgement dropped the Notes column of the English install table, and several of the caveats it carried were load-bearing. Four of seven install rows no longer worked as written: - Homebrew was missing the one-time `brew trust libredb/tap`, which Homebrew 6+ requires before it will install from the tap. - Snap was missing `sudo`. - Helm was `helm repo add`, which installs nothing. Replaced with the OCI one-liner the English README uses. - deb/rpm was merged with the desktop AppImage. These are different artifacts: one is the systemd server package, the other runs the server as a local sidecar with no login prompt. Split, with the English "not the server package" warning restored, and the Flatpak given its own row since it comes from the FlatPark remote rather than the Releases page. Corrected three claims that went further than the English source: - "without an API key this panel does not exist" was false. The NL2SQL and Autopilot tabs render unconditionally; no NEXT_PUBLIC LLM variable exists, so the client cannot know whether a key is configured. The privacy property that is actually true - no key means no call, so nothing leaves the network - is now what the sentence says. - The shared explorer/ER-diagram/schema-diff/monitoring set was claimed across all ten engines. README.md scopes it to the SQL engines and calls out Druid as a double exception; MongoDB and Redis have no ER diagram or schema diff. - Migration SQL generation was unscoped. It covers PostgreSQL, MySQL, SQLite, Oracle and SQL Server, plus ClickHouse column modifications. Restored three details the tables lost against README.md: `transactions` on the Oracle and SQL Server rows, and the leading "none" on the Couchbase, ClickHouse and Druid driver cells, which is where the zero-runtime-dependency position is visible. Added the AUTH_COOKIE_SECURE=false note to both quick starts. Reaching the app over plain HTTP on a LAN makes login fail silently while health checks stay green, and neither file carried an environment table or a link to docs/DISTRIBUTION.md where the fix lives. Japanese fixes: "スマートフォンから届く" inverted the direction of the English "it has to reach a phone"; the specification bullet list mixed plain and polite forms in its last item; `INFER` is column inference, not type inference.
The previous commit reintroduced eleven em dashes, which 1f5f5f6 had just finished removing from these two files. Replaced with the punctuation each sentence actually wanted: parentheses for the parenthetical lists, a semicolon for the Couchbase aside, a full stop before the desktop-app explanation. The driver cells now say "no driver, HTTP only" outright instead of leaning on a dash to carry the contrast.
Member
|
Reviewed and pushed the fixes onto this branch (fd7358a, 22b6115). The translation itself is good — the Chinese reads like it was written for V2EX rather than translated at one, every relative link resolves, and the numbers (ten engines, 8 charts, 5 aggregations, 30+ inferences, 100% coverage, six test layers, the Node 20.9/22/24 smoke matrix, all eleven one-click channels) check out against the English README and What needed fixing came from one root cause: the abridgement dropped the Notes column of the English install table, and several caveats that column carried were load-bearing. Install commands that did not work as written. Homebrew was missing the one-time Three claims that went further than the English source.
Details the tables lost. One addition. Both quick starts now carry the Japanese. Three real ones: Note that 22b6115 exists because my first commit reintroduced eleven em dashes that 1f5f5f6 had just removed. Sorry about the noise. Two things I did not touch, because they are yours to decide:
Separately, and not a defect in this PR: the English engine table lists SSL/TLS and SSH tunnel only on the PostgreSQL and MySQL rows, and both translations faithfully carried that over. |
|
cevheri
approved these changes
Aug 8, 2026
cevheri
added a commit
that referenced
this pull request
Aug 8, 2026
… the transport-security scope (#318) * docs: guard the localized READMEs, state the language policy, correct the transport-security scope Three follow-ups from the #317 review, none of which belonged in that PR. Localized README drift guard. README_zh.md and README_ja.md restate README.md's engine table and install commands, and nothing noticed when they diverged - that is how #317 reached review with a Homebrew row missing its mandatory `brew trust`, a Snap row missing `sudo`, and a Helm row that added a repo without installing anything. `bun run readme:check` enforces two invariants: the engine name set is identical in all three files, and every command in a localized install table appears verbatim in README.md's. Abridgement stays legal - the translations deliberately drop Chocolatey and the portable zip - but paraphrasing a command does not, because a reader copy-pastes it. Tables are located structurally rather than by heading text, since the headings are in Chinese and Japanese. Wired into the required lint-and-build job for the same reason the chart guard lives there: a separate job would be advisory only. Language policy. The translated READMEs invite issues and pull requests in Chinese and Japanese, which was stated nowhere else - CONTRIBUTING.md and the issue and PR templates carried no language policy at all, so the promise was visible only in the two files a maintainer cannot read. CONTRIBUTING.md now states it, and draws the line the repo actually holds: the conversation may be in any of the three languages, but everything that lands in the tree stays in English apart from the translated READMEs themselves. Transport security scope. All three READMEs listed SSL/TLS and SSH tunnel on the PostgreSQL and MySQL rows only, which reads as if the other engines have neither. Neither half was right. The SSH tunnel is opened in factory.ts before createDatabaseProvider and rewrites host and port, so it is provider-independent and works everywhere except SQLite. SSL/TLS is narrower than the dialog implies: postgres, mysql, mssql, couchbase, clickhouse and druid read config.ssl; oracle.ts, mongodb.ts and redis.ts never do. Replaced the per-row claims with one note per file stating both scopes, and recorded the UI half - a panel that accepts a security setting three providers silently drop - as D2 in docs/BACKLOG.md. * docs: narrow the transport-security claim and read only the command column All five Copilot findings on #318 were valid. commandSpans scanned every cell, so README.md's Notes column put `brew update` and `sudo snap logs libredb-studio` into the canonical set. A localized file could then have carried either in its Command cell and passed. It now reads column 1 only, which is where both table shapes keep the command, with a regression test for a note-only span and one for a row that has no such column. The transport-security note overstated both halves, which matters more than usual because the subject is a security setting. The SSH tunnel is provider-independent, but `factory.ts:229` requires a host and a port, and connection string mode clears both in `use-connection-form.ts:173-179`. That mode is offered for MongoDB, Couchbase and ClickHouse, so those connections are not tunnelled even though the panel appears. "Every engine except SQLite" was also wrong on its own terms: the embedded `libredb` provider is file-based too. More importantly, saying Oracle, MongoDB and Redis "connect without SSL/TLS" claimed more than the code supports. What is true is that those three ignore `config.ssl`; the connection may still be encrypted if the connection string asks for it - `oracle.ts:266` passes a supplied string through verbatim, and the MongoDB driver honours `tls=true` in the URI. The defect is a security setting accepted and silently dropped, not guaranteed plaintext. Reworded in all three READMEs and in BACKLOG D2, which now also records the two scope facts so the next reader does not have to re-derive them.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Three channels on the launch plan expect a README in their own language before they will consider a project at all:
So this is a prerequisite, not a nicety.
Why not a full translation
The English README is 830 lines and changes several times a week. A complete mirror would drift, and a stale Chinese README is worse than no Chinese README.
These are 220 lines each, carrying what a reader needs in order to decide: the positioning, quick start, the ten engines, install paths, where the paid line falls, and the test story. The deep sections (development databases, seed connections, deployment recipes) link to the English docs.
Two things that are not translations
Written for that audience specifically rather than carried over:
SCANover key prefixes, neverKEYS *;INFOfor health;SLOWLOG GET/CLIENT LISTfor sessions. That kind of specificity lands well with these readers.Also
The English README gains a language switcher (
English · 简体中文 · 日本語). Without it neither file is reachable.Every relative link in both files was checked against the tree.
docs/,public/screenshots/,LICENSEandCONTRIBUTING.mdall resolve.What I would like reviewed
The Chinese, by someone who reads it. It was written without a native speaker in the loop. It should be technically correct and close to the register used on V2EX and 掘金, but naturalness is exactly what decides whether a HelloGitHub submission is accepted, and that is the one thing this PR cannot verify on its own. The Japanese carries the same caveat with a smaller bet behind it.
Deliberately left out
Neither file links to a live demo.
app.libredb.orgcurrently opens an SSO prompt, and sending this audience there would undo the point. The link goes in once #316 is deployed anddemo.libredb.orgis up.