zhangshenghang opened a new pull request, #12487:
URL: https://github.com/apache/seatunnel/pull/12487
## Purpose of this PR
This PR fixes a batch of real documentation issues found by comparing the
docs on `upstream/dev` against the connector source code, and by auditing
English/Chinese doc parity. All items were verified against the code before
editing.
## What was found and changed
### 1. Broken Doris Source option key (code + docs)
- The async Arrow deserialization queue option was registered under the
malformed key `doris.request.retriesdoris.deserialize.queue.size` — a
concatenation typo introduced in #7895 that glued the preceding option name
(`doris.request.retries`) onto the intended key
(`doris.deserialize.queue.size`). The docs even contained a sentence telling
users to "use this exact key".
- **Change**: `DorisSourceOptions.java` now registers the option as
`doris.deserialize.queue.size`; the corrupted table row was fixed and the
misleading sentence removed in both `docs/en/connectors/source/Doris.md` and
`docs/zh/connectors/source/Doris.md`; an entry was added to
`docs/{en,zh}/introduction/concepts/incompatible-changes.md` per the
backward-compatibility policy (default value `64` and behavior unchanged; only
configs that explicitly used the malformed key are affected).
### 2. Doris Sink: wrong `sink.enable-delete` default
- `DorisSinkOptions.SINK_ENABLE_DELETE` defaults to `false`, but both the
English and Chinese sink docs listed the default as `-`.
- **Change**: default column corrected to `false` in
`docs/{en,zh}/connectors/sink/Doris.md`.
### 3. SelectDB-Cloud: wrong buffer size annotation
- `sink.buffer-size` default `10 * 1024 * 1024` was annotated as "(1MB)" in
both languages, contradicting both the expression and the code
(`DEFAULT_SINK_BUFFER_SIZE = 10 * 1024 * 1024` = 10 MB).
- **Change**: annotation corrected to "(10MB)" in
`docs/{en,zh}/connectors/sink/SelectDB-Cloud.md`.
### 4. Chinese SMB docs incomplete vs English (fallout of #12297, merged
2026-09-26)
- `docs/zh/connectors/source/SmbFile.md` was missing 5 option rows present
in English (`sheet_name`, `xml_row_tag`, `xml_use_attr_format`,
`compress_codec`, `encoding`).
- `docs/zh/connectors/sink/SmbFile.md` was missing 16 option rows present in
English (`custom_filename`, `file_name_expression`, `filename_time_format`,
`filename_extension`, `partition_dir_expression`,
`is_partition_field_write_in_file`, `max_rows_in_memory`, `sheet_name`,
`xml_root_tag`, `xml_row_tag`, `xml_use_attr_format`, `single_file_mode`,
`date_format`, `datetime_format`, `time_format`,
`create_empty_file_when_no_data`), and its multimodal feature description was
missing a sentence.
- Both zh pages also linked the multimodal feature with the anchor
`#multimodal`, which does not resolve on the Chinese site (the zh heading is
`多模态（multimodal）`, anchor `#多模态multimodal` as used by all other zh connector
pages). Fixed to match.
- **Change**: missing rows added (translations of the English rows), anchor
and description completed.
### 5. Chinese `sink-write-modes.md` missing the new SmbFile row
- #12297 added `SmbFile` to the file-sink save-mode support table in English
but not in Chinese.
- **Change**: `docs/zh/connectors/common-options/sink-write-modes.md` table
now matches English.
## Which areas/files were updated
11 documentation files (en + zh) and 1 one-line code change that the doc fix
depends on:
- `seatunnel-connectors-v2/connector-doris/.../DorisSourceOptions.java`
(option key fix)
- `docs/{en,zh}/connectors/source/Doris.md`,
`docs/{en,zh}/connectors/sink/Doris.md`
- `docs/{en,zh}/connectors/sink/SelectDB-Cloud.md`
- `docs/{en,zh}/introduction/concepts/incompatible-changes.md`
- `docs/zh/connectors/source/SmbFile.md`,
`docs/zh/connectors/sink/SmbFile.md`
- `docs/zh/connectors/common-options/sink-write-modes.md`
Both English and Chinese docs were checked and fixed together for every
issue.
## Duplicate check (last 7 days)
Before opening this PR I reviewed all docs-related PRs opened in the last 7
days and the files they touch: #12476 (transforms), #12459 (sink/source/zeta
docs), #12423 (transforms/zeta/connectors), #12409 (IoTDB/HTTP/zh sync), #12378
(connector doc moves incl. FacebookAds/GoogleAds/Salesforce — deliberately NOT
touched here to avoid overlap), #12347 (file connectors/developer docs), #12359
(PostgreSQL CDC audit), #12328. None of these PRs covers the Doris option key
corruption, the SelectDB "(1MB)" annotation, the Doris sink `enable-delete`
default, or the zh SMB doc gaps. No overlap found.
## How this change was verified
- Option tables were cross-checked against the connectors' own
`Options.key(...)` definitions with a scripted comparison; every mismatch in
this PR was confirmed manually in the source (e.g.
`DorisSinkOptions.SINK_ENABLE_DELETE = false`,
`SelectDBSinkOptions.DEFAULT_SINK_BUFFER_SIZE = 10 * 1024 * 1024`, `SmbConf`
runtime checks).
- Relative-link and anchor checker run over `docs/en` and `docs/zh`: the two
broken `#multimodal` anchors are resolved; no new broken links introduced.
- Markdown table column-count check on every edited table: consistent.
- `./mvnw -q -DskipTests -pl seatunnel-connectors-v2/connector-doris
compile` passes, and `./mvnw spotless:apply` on the module produces no changes.
Full `./mvnw verify` was not run because it is a multi-hour build; the only
code change is a string literal in an options class, which the module compile
covers.
- zh SMB option tables now match the English tables exactly (name-by-name
comparison).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
