Mailing-List: contact derby-dev-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: <derby-dev@db.apache.org>
Message-ID: <890247212.1220880288131.JavaMail.jira@brutus>
Date: Mon, 8 Sep 2008 06:24:48 -0700 (PDT)
From: "Kristian Waagan (JIRA)" <jira@apache.org>
To: derby-dev@db.apache.org
Subject: [jira] Commented: (DERBY-3840) The test code executes java
 processes by just executing java instead of using a full path. This may
 cause the wrong java to be picked up.
In-Reply-To: <1019721108.1218727906207.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/DERBY-3840?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12629154#action_12629154 ] 

Kristian Waagan commented on DERBY-3840:
----------------------------------------

Thank you, Kathey.

What I wanted to know about the method, is whether the path is included or not. Your example in the JavaDoc confirms this.
(Nit: Two of the lines have a different indentation than the rest of the JavaDoc).

I just wondered about the system variable for security reasons. Since this is testing code, it doesn't matter I suppose.
I also checked with the Sun VM, and you're allowed to change it there as well. I don't think the VM will use these values internally, but it will report the values set.

> The test code executes java processes by just executing java instead of using a full path. This may cause the wrong java to be picked up.
> -----------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: DERBY-3840
>                 URL: https://issues.apache.org/jira/browse/DERBY-3840
>             Project: Derby
>          Issue Type: Bug
>          Components: Newcomer, Test
>    Affects Versions: 10.5.0.0
>            Reporter: Kathey Marsden
>            Assignee: Kathey Marsden
>         Attachments: DERBY-3840_diff.txt, DERBY-3840_diff2.txt
>
>
> There are places in the code where we execute a java process. e.g.  BaseTestCase.assertExecJavaCmdAsExpected where we execcute a java process and just call "java".  In windows at least the new shell created for the exec'ed process does not share the parent process path but rather picks up the path from the Windows System variables.  This can cause the wrong java to be picked up.  Also the method does not work for j9 because it calls "java" instead of the actual executable name.  Perhaps we should have a method:
>     public static final String getJavaExecutableName() {
> 	    String vmname = getSystemProperty("com.ibm.oti.vm.exe");
> 	    if (vmname == null)
> 	        vmname = getSystemProperty("java.home") + File.pathSeparator + "java";
> 	    return vmname;
> 	}
> There are also issues with this in NetworkServerControlApiTest.  NetworkServerTestSetup and the replication tests and maybe others as well.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.