rwzdk commented on PR #13020:
URL: https://github.com/apache/seatunnel/pull/13020#issuecomment-6503656833

   Thanks — and agreed on the scope: no empty commit, and the checkpoint lane 
stays as it was.
   
   Issues 1–4 and 6 are addressed in `46680e9c` plus the PR description; Issue 
5 stays a follow-up since it is dead private code the change never touches.
   
   **Issue 1 (factory-reachable scenario)** — the reproduction was wrong, as 
you found, and the description now uses a list-valued first path without `[*]`:
   
   ```hocon
   json_field = {
     a = "$.a"       # size 1, routes to JsonPathProcessorImpl
     b = "$.b[*]"    # size 2
   }
   ```
   
   There is a matching test through 
`JsonPathProcessorFactory.getProcessor(paths, true)` that asserts the factory 
returns `JsonPathProcessorImpl` and the padded matrix `[[1,10],[null,20]]`, 
plus one that pins a `[*]` first path to `ArrayJsonPathProcessor` so the 
routing split stays documented.
   
   One correction to the example in your review: `$.id` cannot be the first 
path here. Jayway returns a bare `Integer` for it, and `processJsonData` casts 
each read to `List<String>`, so a scalar path fails with `ClassCastException` 
before and after this change — it is a pre-existing limitation, not something 
the padding change introduces. That is why the factory case uses `$.a`; 
`$..name` would work the same way.
   
   **Issue 2 (assertions)** — cases 1 and 2 now assert the full matrices 
(`[[1,10],[null,20]]` and `[[1,10],[2,null]]`) instead of `assertDoesNotThrow` 
+ row count.
   
   **Issue 3** — `java.util.Arrays` is imported; the factory cases are added.
   
   **Issue 4** — the description now says `json_filed_missed_return_null` 
everywhere (including the config sample); the wrong `json_field_...` spelling 
is gone.
   
   **Issue 6** — one sentence on padding semantics (index-aligned, not 
broadcast) is added to the option table in `docs/en/connectors/source/Http.md` 
and `docs/zh/connectors/source/Http.md`.
   
   Evidence:
   
   - Red: the strengthened tests run against the pre-fix `dataFlip` — `Tests 
run: 6, Failures: 2, Errors: 3`, including `IndexOutOfBoundsException: Index 1 
out of bounds for length 1`.
   - Green: `Tests run: 6, Failures: 0, Errors: 0`.
   - Full `connector-http-base` suite on JDK 17: `Tests run: 74, Failures: 0, 
Errors: 0`.
   
   The push also re-triggered the fork `Build`; I will report if anything comes 
back red. One environment note: on JDK 21 the module's Mockito-based tests fail 
with `Unsupported class file major version 65` before reaching any assertion, 
so run it with JDK 17.
   
   Prepared with an AI coding agent.
   


-- 
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]

Reply via email to