wgzhao commented on PR #12355:
URL: https://github.com/apache/seatunnel/pull/12355#issuecomment-5713342107
Thanks — Issue 1 is addressed in `5507c2c7d`, on top of the `dev` sync you
suggested (`a3115f14`).
**Issue 1 — MODIFY / CHANGE COLUMN coverage.** Added
`testParseAlterTableModifyAndChangeSetAndEnumColumnKeepOptionList` to the
existing test class rather than a new one. It drives a single statement with
three alter options — `MODIFY COLUMN c_set SET('a','b','c') NULL`, `CHANGE
COLUMN c_enum_src c_enum ENUM('x','y') NULL` and `MODIFY COLUMN c_varchar
VARCHAR(64) NULL` — and asserts the rebuilt types `SET('a','b','c')`,
`ENUM('x','y')` and `VARCHAR(64)`, the `STRING` data type on both SET/ENUM
columns, and the old/new names on the CHANGE event.
Two details worth calling out:
- `CHANGE COLUMN` goes through `rename()` before the event is emitted, so we
read `PhysicalColumn.rename` and confirmed it carries `sourceType` across; the
assertion now pins that behaviour.
- The `VARCHAR` option in the same statement is deliberate: it keeps the
fallback to `SeatunnelDDLParser.super` under test, so a future change that
routes every type through the option-list branch fails here.
We also ran these shapes as a scratch test before writing this one (MODIFY
SET, MODIFY ENUM, CHANGE with rename, plus a VARCHAR control) to confirm this
is coverage rather than a latent defect — all four produced the expected types,
so your reading that the shared call path makes it a completeness gap was right.
**Issue 2 — E2E.** We would rather not fold it in here. The only natural
host is `MysqlCDCWithSchemaChangeIT`, a Testcontainers + zeta schema-evolution
suite with ordered cases that already runs under a 300s convergence timeout;
proving "MySQL accepts the generated `SET(...)`" needs a new source/sink
fixture, a job config, an ALTER adding a SET column and a post-DDL assertion —
a piece of e2e work in its own right rather than an extra assertion. If you
agree that a follow-up issue is the right home, we will file it and link it
here; if you would rather see it in this PR, say so and we will add it.
CI is re-running on `5507c2c7d`.
--
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]
