Description Stephan Herrmann 2014-05-01 08:29:07 EDT

This is mostly a test issue, just the last question below might indicate s.t. broken in shippable code.

This file:
   http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/tree/org.eclipse.jdt.core.tests.model/workspace/Converter/src/test0536/A.java

is supposed to be equal to this attachment:
   https://bugs.eclipse.org/bugs/attachment.cgi?id=7629

but in fact the content is duplicated with some extra text in the middle:

Lines 248-260:
-----8<-----
		}
		System.out.println(output);
	
	}
}Content-Type: text/plain

/*
 * Created on Jan 20, 2004
 *
 * To change the template for this generated file go to
 * Window&gt;Preferences&gt;Java&gt;Code Generation&gt;Code and Comments
 */
package test0536;
-----8<-----

Interestingly the Java editor does *not* show anything beyond the brace closing the first (correct) class declaration! You need to use a text editor to see the duplication.

This surfaced by causing a regression in Object Teams where, for unknown reasons, we all the sudden parse the entire file and create a CUD with duplicate content leading to bad AST structure (package declaration coming *after* imports).

Questions:

=> What makes the tool (at parsing level? at input stream level?) stop reading at the closing brace?

=> Has the file been bogus all the time? Is it an artefact of bogus CVS->git conversion?

=> Do we need to handle the case of stray package declaration late in the file? (That's what currently causes the bad AST structure.) Is such handling present in JDT (where?) and just got broken in Object Teams?