[
https://issues.apache.org/jira/browse/DERBY-7037?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Piotr Zygielo updated DERBY-7037:
---------------------------------
Description:
AFTER UPDATE trigger fails when the column of source table is autogenerated.
Bisected issue to fat commit
[a826375a|https://github.com/apache/derby/commit/a826375a5482ff797e90b93a83b12173964fd181]
(DERBY-6414). After that change NPEx is observed and update statement is
forcibly prevented with current connection being closed.
Logs from example executions: [at
Travis|https://travis-ci.org/pzrep/derby_npex_trigger/builds/505226771], with
simple SQLs available on GitHub.
Edit: as checked on [more complex
case|https://travis-ci.org/pzrep/derby_npex_trigger/builds/505297229],
reordering columns is not enough.
was:
AFTER UPDATE trigger fails when the last column of source table is
autogenerated.
Bisected issue to fat commit
[a826375a|https://github.com/apache/derby/commit/a826375a5482ff797e90b93a83b12173964fd181]
(DERBY-6414). After that change NPEx is observed and update statement is
forcibly prevented with current connection being closed.
Logs from example executions: [at
Travis|https://travis-ci.org/pzrep/derby_npex_trigger/builds/505226771], with
simple SQLs available on GitHub.
Not 100% sure about workaround (as I didn't analyse code, only checked some
outcomes), but it seems that reordering columns helps (see [logs for almost the
same SQLs as
above|https://travis-ci.org/pzrep/derby_npex_trigger/builds/505233514]). But
for large production datasets it might not always be feasible to do such
migration.
> NPEx at InternalTriggerExecutionContext.updateAICounters
> --------------------------------------------------------
>
> Key: DERBY-7037
> URL: https://issues.apache.org/jira/browse/DERBY-7037
> Project: Derby
> Issue Type: Bug
> Affects Versions: 10.12.1.1, 10.13.1.1, 10.14.2.0, 10.15.1.3
> Environment: Linux, openjdk 8 & 9
> Reporter: Piotr Zygielo
> Priority: Major
>
> AFTER UPDATE trigger fails when the column of source table is autogenerated.
> Bisected issue to fat commit
> [a826375a|https://github.com/apache/derby/commit/a826375a5482ff797e90b93a83b12173964fd181]
> (DERBY-6414). After that change NPEx is observed and update statement is
> forcibly prevented with current connection being closed.
> Logs from example executions: [at
> Travis|https://travis-ci.org/pzrep/derby_npex_trigger/builds/505226771], with
> simple SQLs available on GitHub.
> Edit: as checked on [more complex
> case|https://travis-ci.org/pzrep/derby_npex_trigger/builds/505297229],
> reordering columns is not enough.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
