Bug 419080 - PlatformObject.getAdapter(Class) can return null for valid adapter due to race condition in AdapterFactoryProxy
Summary: PlatformObject.getAdapter(Class) can return null for valid adapter due to rac...
Status: RESOLVED FIXED
Alias: None
Product: Equinox
Classification: Eclipse Project
Component: Components (show other bugs)
Version: unspecified   Edit
Hardware: PC Windows 7
: P3 major (vote)
Target Milestone: Juno SR2+   Edit
Assignee: John Arthorne CLA
QA Contact:
URL:
Whiteboard:
Keywords: greatbug
Depends on: 408506
Blocks: 405320 409608
  Show dependency tree
 
Reported: 2013-10-09 16:01 EDT by John Arthorne CLA
Modified: 2013-10-09 16:02 EDT (History)
3 users (show)

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description John Arthorne CLA 2013-10-09 16:01:20 EDT 
+++ This bug was initially created as a clone of Bug #408506 +++

We (Dali) adapt the platform interface IProject to the Dali interface ProjectResourceLocator.

This Dali interface and its sole implementation, SimpleProjectResourceLocator, are in the plug-in o.e.jpt.common.core. The IAdapterFactory that implements this "adaptation" is in the same plug-in. And plugin.xml defining the corresponding extension to the o.e.core.runtime.adapters extension point is also in the same plug-in.

The problem: Intermittently, during start-up, a call to IProject.getAdapter(ProjectResourceLocator.class) will return null with the first call but not null with subsequent calls. The adapter factory is implemented in such a way that it will never return null; so, apparently, the adapter factory is not invoked by the platform adapter manager. This seems to be a bug, as the plug-in must have been loaded for the class ProjectResourceLocator to be resolved. Unfortunately, in a test workspace where this happens with every start-up, it never happens when executing under the debugger (even with no breakpoints whatsoever).

Is this expected behavior (i.e. that even though a plug-in is loaded, its adapter factories are not necessarily loaded)? If not, can anyone offer any hints as to how we can investigate, sans the debugger, why the platform adapter manager does not invoke the appropriate adapter factory?