[
https://issues.apache.org/jira/browse/IGNITE-10824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16746334#comment-16746334
]
Pavel Kuznetsov commented on IGNITE-10824:
------------------------------------------
[~vozerov] Thank you,
1. Done
2. Actually we do create light-weight views for columns for the unification. In
{{planForInsert}} we have {{GridSqlColumn[]}}; For updates we've got
ArrayList<GridSqlColumn>; and in planForBulkLoad we've got just array of h2's
{{Column}}s.
3. Ok, I've updated code so, we can avoid the lookup into the hashmap, now we
take in {{#verifyUpdateColumns}} collection (which may be a light-weight view)
of h2's {{Column}}s
4. Sure, done.
> SQL: mixing _key and key columns in the DML queries must be disallowed
> ----------------------------------------------------------------------
>
> Key: IGNITE-10824
> URL: https://issues.apache.org/jira/browse/IGNITE-10824
> Project: Ignite
> Issue Type: Bug
> Components: sql
> Reporter: Pavel Kuznetsov
> Assignee: Pavel Kuznetsov
> Priority: Minor
> Fix For: 2.8
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> DML should contain either placeholder for _key (_val) or subset of key
> (value) columns but not both. Also we should keep in mind _key/_value aliases
> Given table with primary key column {{id}} and value column {{salary}}. Next
> queries should be validated to parsing error:
> {code:sql}
> INSERT INTO TEST_TABLE (_key, id, salary) VALUES (1, 2, 42);
> UPDATE TEST_TABLE SET _val = 1, salary = 2;
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
