SEZ9 commented on PR #12853:
URL: https://github.com/apache/seatunnel/pull/12853#issuecomment-6992022042

   @Eymenonar thanks for the point-by-point confirmation. Going by what you 
describe on `e6985779d2c212f802774aa93840787f3d518a5f`, all three answers 
settle what was asked and need no further code change:
   
   1. **Spark / Iceberg compatibility**: the incompatible-changes entry (EN and 
ZH) now names Spark schema translation and Iceberg table creation, with the 
cast to precision <=38 as the migration path. That was the missing call-out.
   2. **Docs and scale clamp**: the MySQL source and sink mapping tables (EN 
and ZH) now show the same `(p,s)` for signed and unsigned DECIMAL, capped at 
65/30. Both converters cap scale at 30, including after the excess-precision 
reduction, and compute `adjustedScale` once, so the docs and `convert()` now 
agree. Thanks also for clarifying that the table changes live in `e02698b15`, 
not in `7a3c7991` or `825f69f9`.
   3. **Clamp coverage**: `testConvertDecimalClampsPrecisionAndScale` in both 
`MySqlTypeConverterTest` and `OceanBaseMySqlTypeConverterTest`, asserting 
`(70,10) -> (65,5)` and `(50,40) -> (50,30)`, plus `(70,3) -> (65,0)` on the 
MySQL side, covers the previously untested >65 branch.
   
   Three things are still open from the earlier review:
   
   - **OceanBase unsigned rule**: the OceanBase assertions show the old 
`(38,18)` narrowing is gone there, but your comment doesn't say whether 
`OceanBaseMySqlTypeConverter` also dropped the unsigned `p + 1` rule. Could you 
confirm that the same table now yields the same SeaTunnel schema through both 
dialects for `DECIMAL UNSIGNED`?
   - **`testWideDecimalValueIsWrittenWithoutNarrowing`**: this wasn't covered 
in your comment. The concern was that it cannot fail for the behaviour it 
names, because the write path ignores `DecimalType` precision and the literal 
only has 26 digits. Was it reworked, for example with a literal wider than 38 
digits, or renamed or removed?
   - **CI**: as you note, CI on this head is still unverified. Please post here 
once it has run, so we can close this out.
   
   <!-- streview-comment:2432 -->


-- 
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