i18n(ja): fix だ/である plain-style sentences stranded in polite prose - #23584
i18n(ja): fix だ/である plain-style sentences stranded in polite prose#23584yahonda wants to merge 11 commits into
Conversation
Corpus-wide scan for です/ます (polite) vs だ/である (plain/literary)
sentence-register mixing: files that are overwhelmingly polite but
have a stray plain-style sentence or paragraph, which reads jarring
against the surrounding voice. Each site verified against EN and
fixed to match the file's dominant polite register:
- releases/release-6.5.0.md: not just a register issue - the clause
was also word-order scrambled and missing ます entirely
("非トランザクション`INSERT`文`REPLACE`サポートしてい`UPDATE`。"),
garbling which statement types are supported. Rewritten to match
EN's actual list of three statement types.
- develop/java-app-best-practices.md, develop/dev-guide-connection-
parameters.md, information-schema/information-schema-inspection-
result.md, optimistic-transaction.md, grafana-pd-dashboard.md: one
stray plain-ending sentence each, each mismatching its own
immediate sibling sentences/bullets which are polite.
- tidb-lightning/tidb-lightning-physical-import-mode-usage.md: two
plain sentences in the same line fixed together.
- dr-secondary-cluster.md: a whole 4-sentence paragraph was plain-
style, sandwiched between polite text before and after - fixed all
4 sentences together, not just the originally-flagged one.
- pessimistic-transaction.md: two separate sites - a 3-sentence plain
paragraph (fixed as a unit) and one bullet whose ending mismatched
the sibling bullet immediately above it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
|
Important Review skippedDraft detected. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: ASSERTIVE Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
test seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it. |
Second batch from the same corpus-wide register-mixing scan, covering bullet-list items (verified against sibling bullets in the same list, not just isolated prose sentences): - sql-tuning-best-practice.md, online-unsafe-recovery.md: "want to" scenario bullets ending plain (したい。) reworded to polite equivalents matching their sibling bullets. - performance-tuning-methods.md, configure-load-base-split.md, tiproxy/tiproxy-grafana.md: single bullet mismatching its immediate sibling bullets in the same list. - troubleshoot-cpu-issues.md: 2 sites - one bullet with 2 plain sentences, and one whose ending mismatches this same file's other "high load" bullet phrased identically but polite. - tidb-troubleshooting-map.md: 7 sites across a numbered troubleshooting list (4.5.3/4.5.5/4.5.6 and 3 "原因" bullets), confirmed via a sibling item (4.5.4) in the identical list that already uses polite form - including one multi-sentence RocksDB bullet where all 5 sentences were plain, fixed together. - releases/release-5.0.0.md: isolated plain sentence in an otherwise fully polite file. - optimistic-transaction.md: also fixed a stray duplicate closing "。" (space + period) found by the user while reviewing the previous commit's context, unrelated to the register issue but on the same already-touched paragraph. Also flagging (not fixed): tidb-performance-tuning-config.md and clinic/clinic-data-instruction-for-tiup.md had a similarly-shaped candidate that turned out to be a different defect - a list item became a full verb clause where all its sibling items (and EN) are bare noun phrases - a parallelism mismatch, not register mixing. daily-check.md's flagged bullets turned out to be an internally consistent plain-style "reasons" list matching EN's own terse fragments - not a defect. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Found while investigating style-register candidates that turned out
to be a different defect: these 2 list items became full verb
clauses ("〜が多い。"/"〜が遅い。") while every sibling item in the
same list (and EN) is a bare noun phrase. Fixed to match:
- tidb-performance-tuning-config.md: "同時接続数が多い。" (the number
of concurrent connections is high) -> "多数の同時接続。" (many
concurrent connections), matching EN "High concurrent connections."
and sibling items "列数の多い大きな表。"/"複雑なSQLクエリ。"/
"多様なクエリパターン。"
- clinic/clinic-data-instruction-for-tiup.md: "クエリ ログ ファイルが
遅い。" (the query log file is slow) -> "スロークエリログファイル。"
(slow query log file), matching EN "Slow query log files." and this
corpus's established term スロークエリログ (128 occurrences), and
the sibling bullets' "...ログファイル。" noun-phrase structure.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Extended the corpus-wide register-mixing scan to also catch bare
dictionary-form verb endings (表示する。/記録する。/占める。 etc.),
which the earlier passes deliberately excluded as too noisy. Filtered
157 raw candidate files down to a manageable set by requiring 3+
plain-ending sentences per file, then verified each against EN and
sibling list items before fixing:
- check-before-deployment.md: numbered steps 1 and 4 ("...を表示する。")
mismatch their sibling steps 2/3/5/6/7, which are all polite.
- scale-tidb-using-tiup.md: 5 numbered-step sites, same "...を表示する。"
pattern, confirmed against sibling steps in the same procedures.
- develop/dev-guide-sample-application-python-django.md: a 6-item
feature bullet list where only the first item was polite and all 5
remaining items were plain - fixed all 5 to match.
- system-variables.md: 4 more sites beyond the one already fixed on
the fix-mega-files-review-2026-08-20 branch - a Note blockquote's
closing sentence, one variable's action-description bullet
(established convention in this file is always polite for these),
one usage-example's opening imperative sentence mismatching its own
second sentence, and one "scenario" condition duplicated at 2
separate variable entries (both instances fixed).
This scan is broader but still not exhaustive - many more files with
1-2 plain outliers exist below the 3+ threshold used here to keep
scope manageable; see memory for the full method and remaining gaps.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Batch 4 of the ongoing register-mixing sweep. 5 sibling
dev-guide-sample-application-*.md files (java-jdbc, python-mysqlclient,
golang-sql-driver, python-mysql-connector, python-pymysql) share an
identical "Are you using a driver or an ORM framework?" section with
the same defect repeated in all 5: the intro sentence and the 2nd of
3 bullets end plain ("...となる。"/"...管理する。") while bullets 1
and 3 in the identical list are polite ("...します。"). Fixed both
sites in all 5 files.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
9 more files from the ongoing register-mixing sweep, each verified
against sibling list items or surrounding prose in the same file:
- replicate-between-primary-and-secondary-clusters.md: 3 sites (1
intro-list item, 2 numbered-step titles "GCを無効/有効にする。").
- read-historical-data.md: 2 numbered-step titles mismatching the
other 5 steps in the same list.
- tidb-cloud/get-started-with-cli.md: 2 identical instructional
sentences ("View all commands available:") before code blocks.
- sql-statements/sql-statement-explain-analyze.md: a run-on
lettered sub-step list (a/b/c) all plain while its own outer
numbered steps 1 and 3 are polite.
- alert-rules.md: 2 sites in "解決:" blocks, mismatching the other
12 of 14 such blocks in this file.
- ticdc/ticdc-overview.md: a capability bullet + a prose sentence
introducing a diagram.
- tikv-in-memory-engine.md: 2 sites, one bullet and one numbered
step with 2 plain sentences.
- ticdc/ticdc-sink-to-mysql.md: 2 of 3 items in one bullet list.
- ai/integrations/tidb-mcp-server.md: 1 prose sentence + 1 bullet.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
10 more files, 20 sites, from the ongoing register-mixing sweep, focused on release-notes and feature/use-case bullet lists. Each site verified against sibling bullets in the identical list (which are polite です/ます) before fixing: - tidb-cloud/releases/release-notes-2024.md: 3 separate top-level bullet-list outliers under different date headings. - releases/release-5.2.0.md, release-5.1.0.md, release-6.2.0.md, release-6.4.0.md, release-5.4.0.md: top-level "main new features" or use-case list items mismatching their siblings. - tidb-cloud/backup-and-restore.md and backup-and-restore-concepts.md share an identical duplicated 3-item use-case list template with the same 2 defects in both files. - tidb-cloud/security-concepts.md: 4 sites across separate use-case/ example bullets, each mismatching a nearby polite sibling. - tidb-cloud/database-schema-concepts.md: 2 sites, lower-confidence (both plain sentences agree with each other with no polite sibling in the same short list, but mismatch the surrounding polite prose). Not fixed (explicitly flagged as false positives): release-5.2.0.md line 214 and release-5.4.0.md line 320 sit inside entrenched, internally-consistent mixed-style per-PR changelog sub-lists that are too widespread to fix piecemeal without a dedicated pass. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
"[REPEATABLE READ`分離](link)破損" read as a garbled compound noun "REPEATABLE READ separation-corruption" with no connector. EN: "damaged REPEATABLE READ isolation" - added の so it reads "REPEATABLE READ分離の破損" (corruption OF REPEATABLE READ isolation), found by the user while re-reading this already-touched paragraph. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Final batch of the register-mixing sweep: ~50 files with a single flagged plain-ending sentence each, each independently verified against sibling sentences/bullets/steps in the same list or paragraph before fixing. Includes several shared-template defects duplicated across sibling files: - changefeed-sink-to-apache-kafka.md / -pulsar.md / -mysql.md / migrate-from-mysql-using-data-migration.md: identical "Accepter DNS resolution" step (4 files, same fix). - migrate-from-op-tidb.md / migrate-from-op-tidb-premium.md: identical "環境を構築し、ツールを準備する" step (2 files). - migrate-from-csv-files-to-tidb.md / migrate-from-parquet-files-to- tidb.md: identical "方法2:対象テーブルのスキーマを手動で作成する" bullet (2 files). Also, separately, a corpus-wide sweep for "[バグを報告する](link)。" (report-a-bug link used as a bare dictionary-form predicate, ending sentences with no polite auxiliary) found the same defect repeated 9 times in tidb-troubleshooting-map.md, 3 times in troubleshoot-cpu-issues.md, and a related missing-こと grammar defect in troubleshoot-tidb-cluster.md (3 sites: "...できます" attached directly to a dictionary-form verb instead of "...することができます", and a stray half-width "." instead of "。" in two spots). Two of the tidb-troubleshooting-map.md instances were also scrambled/garbled sentences beyond just the register issue (lines describing TiKV panic recovery and a GC-disable-then-report-bug procedure) - reconstructed to match EN's actual meaning and order. Not fixed (explicitly flagged as out of scope by the verification passes): several sites turned out to be headings/frontmatter/table- captions/definition-list items with no real polite sibling to contradict (these are a file's own consistent convention, not a register mixing defect), a few sit inside entrenched inconsistently-styled release-notes changelog sub-lists too widespread to fix piecemeal, and 2-3 sites are a different defect class (noun-phrase-vs-clause parallelism or an EN-inherited structural issue) flagged for a separate pass. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Corrects a mistake introduced in the previous batch: the intro
sentence "移行タスクを作成するには、以下のいずれかの方法があります。"
(there are the following methods) introduces a list of METHODS, but
both bullets were phrased as imperative commands
("...してください"). EN confirms these are method-description
phrases ("By following the WebUI instruction." / "By using a
configuration file."), not instructions to follow right now. Changed
both to "〜する方法。" (the method of doing X), matching EN and
flowing naturally from the intro sentence. Found by the user
re-reading the rendered text.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Revised dm-webui-guide.md's 2 method bullets from "〜する方法。" (noun
phrase) to "〜します。" (plain declarative) per user feedback - reads
more naturally as a description than a nominalized fragment.
Searched the corpus for the same defect shape: a bullet enumerating
one of several alternative methods (following an intro like "以下の
いずれかの方法で〜できます") phrased as an imperative
("してください") instead of matching its sibling bullets' declarative
style. Found and fixed 3 more:
- dashboard/top-sql.md: sole access-method bullet was imperative
("クリックしてください。") while the intro says "アクセスできます"
(can access, not "please do") - changed to declarative
"クリックします。"
- migrate-from-parquet-files-to-tidb.md: the "check progress via
monitoring dashboard" bullet was imperative, while its 3 sibling
files (migrate-large-mysql-to-tidb.md, migrate-aurora-to-tidb.md,
migrate-from-csv-files-to-tidb.md) all correctly use declarative
"確認します。" for the identical bullet - fixed to match.
- tidb-lightning/tidb-lightning-distributed-import.md: same bullet
mismatched its own sibling bullet immediately above it in the same
2-item list (which is declarative).
Checked ~65 other "as one of the following methods" intro sentences
across the corpus for the same imperative-vs-declarative mismatch;
these 3 (plus dm-webui-guide.md) were the only genuine hits - most
other files already use consistent declarative bullets.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
What is changed, added or deleted? (Required)
Corpus-wide scan for です/ます (polite) vs だ/である (plain/literary) sentence-register mixing: files that are overwhelmingly polite but have a stray plain-style sentence or paragraph, which reads jarring against the surrounding voice. Each site verified against EN and fixed to match the file's dominant polite register (9 files):
releases/release-6.5.0.md: not just a register issue - the clause was also word-order scrambled and missing ます entirely, garbling which statement types are supported. Rewritten to match EN.develop/java-app-best-practices.md,develop/dev-guide-connection-parameters.md,information-schema/information-schema-inspection-result.md,optimistic-transaction.md,grafana-pd-dashboard.md: one stray plain-ending sentence each, mismatching immediate sibling sentences/bullets.tidb-lightning/tidb-lightning-physical-import-mode-usage.md: two plain sentences in the same line fixed together.dr-secondary-cluster.md: a whole 4-sentence paragraph was plain-style, sandwiched between polite text - fixed all 4 sentences together.pessimistic-transaction.md: a 3-sentence plain paragraph plus one bullet mismatching its sibling above.This is the first batch from an ongoing scan; more candidates (bullet-list/table-cell style, which need per-list context judgment) are still being verified and may follow as additional commits on this PR.
Which TiDB version(s) do your changes apply to? (Required)
What is the related PR or file link(s)?
AI agent involvement
Do your changes match any of the following descriptions?