docs(gaussdb): add the missing Supported Data Types section - #233
docs(gaussdb): add the missing Supported Data Types section#233heywalter wants to merge 2 commits into
Conversation
The connector declares 66 data types; this page had no type table at all. 64 of them are listed here, grouped using the same category vocabulary as the other PostgreSQL-family pages (postgresql.md, opengauss.md). Two spec entries are deliberately left out — see the PR description. Evidence: connectors/huawei-cloud-gaussdb-connector/src/main/resources/temp.json tapdata-connectors @ 76fbafd39a2739ab847eb12fb0c38511eade112f DocsLoop-Task: T-gaussdb
heywalter
added
the
docs-loop
✅ Deploy Preview for tapdata-en ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 30dcb0c12c
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Addresses the Codex review: the table showed a bare TIME while spelling out both TIMESTAMP variants, so readers would reasonably conclude TIME WITH TIME ZONE is unsupported. Root cause was on our side: the type-name normaliser treated all bracketed parts as parameter placeholders and stripped them, collapsing TIME [($fraction)] [WITHOUT TIME ZONE] TIME [($fraction)] [WITH TIME ZONE] into a single TIME. TIMESTAMP survived only because its time-zone suffix happens to sit outside the brackets in the same spec. Now brackets containing $ are treated as placeholders and dropped; brackets without $ are part of the type name and kept.
|
Good catch — confirmed and fixed in the latest commit. You were right that both variants exist in the connector metadata. The root cause was on our side: our type-name normaliser treated every bracketed segment as a parameter placeholder and stripped it, so both collapsed to Fixed by distinguishing the two uses of brackets: segments containing
Regression tests added so the two variants can't collapse again. |
1 participant
What
Adds the Supported Data Types section, which this page was missing entirely.
The connector declares 66 types; 64 are now listed, grouped into 11 categories.
Section placement follows the ordering used by sibling pages — right after
Supported Versions, same as
postgresql.mdandopengauss.md.How the categories were chosen
Categories are not invented and not taken from the spec's internal
tofield (which maps everything toTapString/TapNumber/… — accuratefor the mapping engine, meaningless to a reader:
uuid,inetandxmlwould all land in one row).
Instead they are learned from the already-published connector pages, so this
table uses the same vocabulary readers already see elsewhere. Where a type
appears in no existing page, it was classified separately and confirmed by a
human before being used.
Two types deliberately omitted
NTERVAL [$FIELDS] [($fraction)]INTERVAL— the leadingIis missing. Sibling connectors (postgres, opengauss) spell itinterval.precisiondouble precisionbeing split.Both look like defects in the connector spec rather than real types.
Documenting them would carry a typo into user-facing docs where no reader
could tell it was one. Suggest fixing them in
temp.jsoninstead; happy to add them here once corrected.Evidence
Every type in the table comes from this file, pinned to the commit that was read:
temp.json— tapdata-connectors @76fbafdValidation
origin/main(baseline already carries 83 pre-existing warnings — the delta is what matters)
🤖 Proposed by Docs Loop ·
DocsLoop-Task: T-gaussdbReply on any line if a call is wrong — we tighten the rule, not just this PR.