[ 
http://jira.codehaus.org/browse/MEXEC-68?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=179887#action_179887
 ] 

Trygve Laugstol commented on MEXEC-68:
--------------------------------------

I was not asking why it is a good idea, I was wondering why the existing 
functionality should be kept around. What good is it when the manifest method 
is implemented?

> classpath too long problem
> --------------------------
>
>                 Key: MEXEC-68
>                 URL: http://jira.codehaus.org/browse/MEXEC-68
>             Project: Maven 2.x Exec Plugin
>          Issue Type: Bug
>          Components: exec
>    Affects Versions: 1.1.1
>         Environment: windows xp, java 1.6
>            Reporter: Mathias Arens
>         Attachments: exec-maven-plugin-long-classpath.patch
>
>
> If you try to run a project main class with a long list of dependencies the 
> classpath grows larger than windows can handle on its command line. The maven 
> exec plugin simply fails when the classpath is getting to large.
> The surefire plugin solves this issue by putting the classpath and the main 
> class into a Manifest.mf file and wrapping it into a jar. I copied the 
> createJar() routine from the ForkConfiguration class from the surefire-booter 
> module to the ExecMojo plus two additional classes.
> I introduced a new attribute to the ExecMojo called longClasspath that can be 
> set from the command line with -Dexec.longClasspath=true. If this flag is set 
> to true a temporary jar file is created and executed.  The command line 
> arguments '-classpath %classpath mainClass' are replaced by '-jar 
> maven-exec.jar' whilst the jar contains the main class and the classpath.
> I added a patch with my current implementation of the ExecMojo class based on 
> revision 9948. The patch works perfectly fine even with maven modules that 
> have long classpaths and where main classes cannot be executed with the 
> current exec-plugin. But in the implementation I made several assumptions on 
> the ordering of the command line arguments. I don't know all the assumptions 
> are valid. At least the patch is a good starting point for a fix of this 
> issue.
> Maybe the default value of the longClasspath attribute should be set to true 
> in further releases because this is the default in the surefire plugin as 
> well. And maybe somebody has a better attribute name for the longClasspath 
> attribute.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: 
http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


Reply via email to