Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 49557

Summary:

[JUnit] PDE JUnit is using internal RemoteTestRunner

Product:

[Eclipse Project] JDT

Reporter:

Wassim Melhem <wassim.melhem>

Component:

Assignee:

Markus Keller <markus.kell.r>

Status:

VERIFIED FIXED

QA Contact:

Severity:

normal

Priority:

CC:

benno.baumgartner, martinae, ryanman

Version:

3.2

Flags:

markus.kell.r: review+

Target Milestone:

3.4 RC1

Hardware:

OS:

Windows XP

Whiteboard:

Attachments:

Description	Flags
patch	none

Description Wassim Melhem

2004-01-05 23:12:45 EST

PDE got a friendly reminder from the API police that we are using the 
following internal classes from JDT JUnit:

org.eclipse.jdt.internal.junit.runner.RemoteTestRunner
org.eclipse.jdt.internal.junit.launcher.JUnitBaseLaunchConfiguration
org.eclipse.jdt.internal.junit.launcher.JUnitLaunchShortcut
org.eclipse.jdt.internal.junit.launcher.JUnitMainTab

We subclass the first three classes and use the fourth one as-is.
Please make them API, as the need for them is not going away.

Comment 1 Wassim Melhem

2004-04-21 13:08:31 EDT

Erich, is this one going to happen in 3.0?

Comment 2 Erich Gamma

2004-04-21 13:37:04 EDT

no the benefit is small and this would just cause another ripple and we want 
to stop the ripples now.

Comment 3 Erich Gamma

2006-05-24 05:48:27 EDT

Moving back to the JDT/UI inbox

Comment 4 Markus Keller

2006-09-15 12:45:59 EDT

Parts of this has already been done in HEAD, but we need to clean up JUnitBaseLaunchConfiguration (and it's PDE-subclass) first to continue:
- decide what attributes should really be visible/changeable for subclasses
- maybe use a separate class to gather all the info that's currently in createVMRunner's arguments
- rename to *Delegate; the current name is very confusing

Comment 5 Markus Keller

2007-04-17 13:37:05 EDT

All internal dependencies except for RemoteTestRunner have been removed.
Moving to 3.4 to see if or when we want to resolve this.

Comment 6 Wassim Melhem

2007-04-17 23:03:43 EDT

Let's hope it's a question of when/how, not if.

The number of discouraged accesses in the SDK has to go down (ultimately to 0) at some point.

Comment 7 Markus Keller

2007-04-18 06:47:12 EDT

I agree. We just have to make sure that the new API does not expose the internal protocol between the host and the target VM

Comment 8 Martin Aeschlimann

2008-05-02 10:28:24 EDT

not for 3.4.

Comment 9 Wassim Melhem

2008-05-02 10:39:06 EDT

Since we want to be API-clean, we have no choice but to remove JUnit plug-in testing from the SDK and have it as an optional download on the PDE page.  How do you like that?

Comment 10 Martin Aeschlimann

2008-05-02 11:44:47 EDT

Created attachment 98455 [details]
patch

This patch will solve the horrible dilemma we're in.

Comment 11 Wassim Melhem

2008-05-02 12:29:00 EDT

Martin, ideally friends should not be crossing feature boundaries.  I don't think PDE should get a special treatment from PDE.

I would prefer to keep the bug open.  It is actually kind of fun to bug you every few months :)

Comment 12 Martin Aeschlimann

2008-05-09 05:23:38 EDT

Markus, can you review the patch in comment 10?

Comment 13 Markus Keller

2008-05-09 10:50:08 EDT

+1. The patch solves the horrible dilemma.

However, this does not solve all friendship problems between jdt.junit and pde.ui:
Bug 216946 has recently added an 'unfriendly' access in org.eclipse.pde.ui.launcher.JUnitLaunchConfigurationDelegate .

Comment 14 Markus Keller

2008-05-13 10:42:42 EDT

Released patch to HEAD.
The other problem from comment 13 could eventually be handled in another bug.

Comment 15 Benno Baumgartner

2008-05-19 05:46:23 EDT

verifying...

Comment 16 Benno Baumgartner

2008-05-19 05:49:36 EDT

verified that we have a new friend in I20080516-1333