[jboss-jira] [JBoss JIRA] (JBJCA-1321) Statement.cancel() is not invoked until the statement is completed

[lorenzo benvenuti (JIRA)]

    [ https://issues.jboss.org/browse/JBJCA-1321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13303305#comment-13303305 ] 

lorenzo benvenuti commented on JBJCA-1321:
------------------------------------------

Hi, 

unfortunately this bug has became critical for our customers and we can't wait for a new version of Wildfly. We were thinking to proceed in this way:

- Download sources for tag 1.2.5.Final (the version Wildfly 9.0.2 is using)
- Apply patches from the PRs
- Build a new jar and swap the original with it

Should we perform some additional operation in order to be able to cancel the statement (such as enabling it in some configuration file or something like that)? Judging from the PRs I guess overwriting the jar should be enough but we want to be sure.

Thank you,

lorenzo

> Statement.cancel() is not invoked until the statement is completed
> ------------------------------------------------------------------
>
>                 Key: JBJCA-1321
>                 URL: https://issues.jboss.org/browse/JBJCA-1321
>             Project: IronJacamar
>          Issue Type: Bug
>          Components: JDBC
>    Affects Versions: WildFly/IronJacamar 1.3.4.Final, 1.2.7.Final
>            Reporter: lorenzo benvenuti
>            Assignee: Lin Gao
>            Priority: Critical
>             Fix For: WildFly/IronJacamar 1.3.5.Final, 1.2.8.Final
>
>
> Hi,
> in our application we are using the {{Statement.cancel()}} method to stop long-running queries; in Wildfly 9.0.2 this is not working because the {{cancel()}} method is synchronized using a lock which is not released until the query is executed. In {{WrappedStatement}}:
> {code:java}
> public void cancel() throws SQLException
>     {
>       if (doLocking)
>           lock();
>        try
>        {
>         /* ... */
> {code}
> It seems this behaviour has changed from version 1.2.5.Final of ironjacamar-jdbc; in version 1.2.4.Final {{WrappedStatement.cancel}} doesn't try to obtain the lock. 
> Probably I'm missing something, but to me it's strange that in order to cancel a statement you have to wait for its completion.
> Thank you,
> lorenzo



--
This message was sent by Atlassian JIRA
(v6.4.11#64026)