DanielLeens commented on PR #10453:
URL: https://github.com/apache/seatunnel/pull/10453#issuecomment-5701052658
Thanks for the fast turnaround on all three points, @wgzhao.
I checked #12333 to confirm before replying here:
- Issue 1 (root cause): confirmed,
`JdbcCatalogUtils.isNumericUnsignedColumnType(...)` now derives `unsigned` from
the leading type-name token instead of a whole-`COLUMN_TYPE` substring match,
and it's wired into both `MySqlCatalog.buildColumn` and
`OceanBaseMySqlCatalog.buildColumn`. That's exactly the substring-match bug
SEZ9 and I both flagged, fixed at its actual source.
- Issue 2 (E2E): confirmed too —
`JdbcMysqlIT.testSetAndEnumColumnWithUnsignedWordInValueList()` in #12333
creates a real `SET('REAL_AS_FLOAT','NO_UNSIGNED_SUBTRACTION')` /
`ENUM('unsigned','other')` table, drives real catalog discovery, and asserts
both map to `STRING` while `INT UNSIGNED` still correctly widens to `LONG`.
That's the real end-to-end proof my prior round asked for, so I'm satisfied
this PR's own unit-test-only coverage for the defensive branch is fine as-is —
the two together give complete coverage.
- Issue 3: agreed this is pre-existing and out of scope for both PRs. I'd
say yes, a tracking issue is worth filing — a CDC-sourced `SET`/`ENUM` column
producing a bare type name instead of the full value list will bite someone
using auto-create-table on a MySQL sink eventually, and it's better to have it
recorded than rediscovered from scratch.
I'll pick up #12333 on its own thread. On this PR, the head is still
`04abfb922` (matches my last review), and the required `Build` check is still
red/`mergeStateStatus: BLOCKED` as of now — that's the same environmental
Postgres/openGauss CDC E2E flake I diagnosed last round, unrelated to this
diff. Nothing new needed from you here; once that clears (or the failed jobs
are rerun), this is still ready to merge from my side.
--
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]
