program-repair / defects4j-dissection Goto Github PK

According to the information given below and the detailed list shown in program-repair.org/defects4j-dissection/#!/ , it seems that there is no difference between Closure 92 and 93...

I am not sure whether I misunderstand the two bugs ... Any help would be much appreciated.

In Junit:

• A test failure is a test that does not pass due to an exception that extends AssertionError (or maybe Error, to be verified)
• A test error is a test that does not pass due to an unexpected exception that does not Extend AssertionError.

In order to be easier for people to select the ones containing failures and/or errors, what about adding a filter on the test problem (error vs failure)?

(My assumption is that it's a different task, probably easier, to not throw an exception than to satisfy a set of assertions)

Remember to include an option to select alternatives views of the patch. Beyond the merged view (current implementation), it is interesting to include a view where the buggy and fixed source code are in parallel columns.

Hi,

Any chance you could share the script / program you used to extract the repair actions in a D4J patch? I.e., the script you used to generate the defects4j-bugs.json file.

Thanks in advance.

For example, given the diff result of two java file /b_OrganizeImportsOperation.java and f_OrganizeImportsOperation.java.

32d31
< import org.eclipse.core.resources.ResourcesPlugin;
557c556
< 		return ResourcesPlugin.getWorkspace().getRoot();
---
> 		return fCompilationUnit.getResource();

What the pattern of this bug? That is, mcRepl(Method call replacement) and retExpChange (Return expression modification) , which one is better?

Thanks!