github-actions[bot] commented on code in PR #64850:
URL: https://github.com/apache/doris/pull/64850#discussion_r3473353065
##########
fs_brokers/cdc_client/src/main/java/org/apache/doris/cdcclient/source/deserialize/PostgresDebeziumJsonDeserializer.java:
##########
@@ -223,45 +179,46 @@ public DeserializeResult deserialize(Map<String, String>
context, SourceRecord r
db, resolveTargetTable(tableId.table()), colName));
}
- for (Column col : pgAdded) {
+ for (Column col : added) {
if (excludedCols.contains(col.name())) {
- // The column is excluded from sync — Doris table does not
have it,
- // so skip the ADD DDL.
- // case: An excluded column was dropped and then re-added.
LOG.info(
- "[SCHEMA-CHANGE] Table {}: added column '{}' is
excluded from sync,"
- + " skipping ADD DDL",
+ "[SCHEMA-CHANGE] Table {}: added column '{}' is
excluded, skipping ADD",
tableId.identifier(),
col.name());
continue;
}
String colType = SchemaChangeHelper.columnToDorisType(col);
- String nullable = col.isOptional() ? "" : " NOT NULL";
- // pgAdded only contains columns present in afterSchema, so field
lookup is safe.
- // afterSchema.defaultValue() returns an already-deserialized Java
object
- // (e.g. String "hello", Integer 42) — no PG SQL cast suffix to
strip.
- // PG WAL DML records do not carry column comment metadata.
- Object defaultObj =
afterSchema.field(col.name()).schema().defaultValue();
+ String defaultValue =
stripPgDefault(col.defaultValueExpression().orElse(null));
+ String nullable;
+ if (!col.isOptional() && defaultValue == null) {
+ // NOT NULL without a usable default would fail ADD COLUMN
against existing Doris
Review Comment:
This fallback silently changes the result for PostgreSQL columns that have a
DEFAULT expression we do not understand. For example the new test case uses
`date NOT NULL DEFAULT current_date`: PostgreSQL backfills existing rows with
the statement date, but `stripPgDefault` returns null, this branch drops `NOT
NULL`, emits `ADD COLUMN d DATE` with no DEFAULT, and then `updatedSchemas`
advances to the fresh PG schema. The rows already loaded into Doris will remain
NULL while PG has non-null default values, and future retries will not revisit
the DDL because the baseline has advanced. Please either map the default safely
or fail/skip automatic DDL without advancing the baseline so this case requires
manual handling instead of silent divergence.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
