yujun777 commented on code in PR #68646:
URL: https://github.com/apache/doris/pull/68646#discussion_r4140900360
##########
fe/fe-core/src/main/java/org/apache/doris/alter/Alter.java:
##########
@@ -373,8 +374,30 @@ private boolean
processAlterOlapTableInternal(List<AlterOp> alterOps, OlapTable
throw new DdlException("Invalid alter operations: " +
currentAlterOps);
}
if (needChangeMTMVState(alterOps)) {
- Env.getCurrentEnv().getMtmvService()
- .alterTable(oldBaseTableInfo, newBaseTableInfo,
currentAlterOps.hasReplaceTableOp());
+ // Which columns an operation's effect on a view turns on is the
operation's to say, see
+ // AlterOp#queryJudgedColumnNames, and every clause of the alter
has to name them: a batch that
+ // mixes a dropped column with a type change is decided by neither
-- no query says anything
+ // about a type change -- and stays invalidated the way it was
before the queries were asked at
+ // all. Each of them also has to have reached the table. A schema
change that is not a light one
+ // is applied by a job, which may not have run where this hook
runs: the table still holds the
+ // column the change takes away, every query still analyses
against it, and an invalidation
+ // decided on that answer would be about the table from before the
change. What is asked is
+ // whether the change has reached the table, which is the same
fact the re-analysis reads, so
+ // the two answers cannot disagree.
+ boolean judgedByQuery = alterOps.stream().allMatch(op ->
!op.queryJudgedColumnNames().isEmpty()
Review Comment:
Coverage for that answer, added after the reply above (25ba2d585ec): the
drop suite now has a case for each of the two ways an add is written, on tables
whose changes a job applies -- `ADD COLUMN spare` and `ADD COLUMN (spare,
other)` -- both with a name no query reads. Each is invalidated, and each fails
when its own operation falls back to the default answer, so the two overrides
are pinned separately rather than through each other.
--
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]
