Recent changes to bugs

Hansel 2 & Junit 4

2007-02-08T18:21:13Z

Trying to follow the examples with setting up Hansel 2 based JUnit 4
tests and I am not able to get the code to execute.

I have a simple test:

package com.eds.junit;

import org.hansel.CoverageRunner;

import static org.junit.Assert.assertTrue;
import org.junit.Test;
import org.junit.runner.JUnitCore;
import org.junit.runner.RunWith;
import org.junit.runners.Suite;

@RunWith(CoverageRunner.class)
@Suite.SuiteClasses({ SomeTest.class })
@CoverageRunner.CoverClasses({ Some.class })
public class SomeTest
{
public SomeTest()
{
}

@Test
public void xyz()
{
Some some = new Some();
assertTrue("Test", true);
}

public static void main(String... arguments)
{
JUnitCore core = new JUnitCore();
//core.addListener(new SomeListener());
core.run(SomeTest.class);
}

}

That tests an even simpler object:

package com.eds.junit;

public class Some
{
public Some()
{
}
}

When I execute this code following the steps in the tutorial I get
errors, the code works fine without Hansel. We have looked into Hansel
1.x and got proof of concepts working with that, now since we are
switching to Java 1.5 we want to start looking at Hansel 2.

The results that I'm getting are:

D:\>java -classpath "D:\Program
Files\Oracle\JDeveloperStudio10.1.3.2\jdev\mywor
k\JUnit4Application\JUnit4Project\classes;D:\Tools\junit4.1\junit-4.1.ja
r;D:\Too
ls\hansel-2.0a3\lib\asm-all-3.0_RC1.jar;D:\Tools\hansel-2.0a3\lib\hansel
.jar" -j
avaagent:D:/Tools/hansel-2.0a3/lib/hansel.jar com.eds.junit.SomeTest
*** java.lang.instrument ASSERTION FAILED ***: "!errorOutstanding" with
message
transform method call failed at
../../../src/share/instrument/JPLISAgent.c line:
831
*** java.lang.instrument ASSERTION FAILED ***: "!errorOutstanding" with
message
transform method call failed at
../../../src/share/instrument/JPLISAgent.c line:
831
*** java.lang.instrument ASSERTION FAILED ***: "!errorOutstanding" with
message
transform method call failed at
../../../src/share/instrument/JPLISAgent.c line:
831
*** java.lang.instrument ASSERTION FAILED ***: "!errorOutstanding" with
message
transform method call failed at
../../../src/share/instrument/JPLISAgent.c line:
831
*** java.lang.instrument ASSERTION FAILED ***: "!errorOutstanding" with
message
transform method call failed at
../../../src/share/instrument/JPLISAgent.c line:
831
*** java.lang.instrument ASSERTION FAILED ***: "!errorOutstanding" with
message
transform method call failed at
../../../src/share/instrument/JPLISAgent.c line:
831

This Assertion Failed error continues on until IllegalState exceptions
start to get thrown:

java.lang.IllegalStateException: Inconsistent instrumentation: 9 != 8
at
org.hansel.HanselCodeAdapter.visitEnd(HanselCodeAdapter.java:312)
at org.objectweb.asm.tree.MethodNode.accept(Unknown Source)
at org.objectweb.asm.tree.MethodNode.accept(Unknown Source)
at org.objectweb.asm.tree.ClassNode.accept(Unknown Source)
at org.hansel.Transformer.transform(Transformer.java:47)
at sun.instrument.TransformerManager.transform(Unknown Source)
at sun.instrument.InstrumentationImpl.transform(Unknown Source)
at sun.instrument.InstrumentationImpl.redefineClasses0(Native
Method)
at sun.instrument.InstrumentationImpl.redefineClasses(Unknown
Source)
at org.hansel.Startup.init(Startup.java:97)
at org.hansel.CoverageRunner.init(CoverageRunner.java:78)
at
org.hansel.CoverageRunner.getDescription(CoverageRunner.java:116)
at
org.junit.internal.runners.CompositeRunner.getDescription(CompositeRu
nner.java:36)
at
org.junit.internal.runners.TestClassRunner.getDescription(TestClassRu
nner.java:57)
at
org.hansel.CoverageRunner.getDescription(CoverageRunner.java:119)
at
org.junit.internal.runners.CompositeRunner.getDescription(CompositeRu
nner.java:36)
at
org.junit.internal.runners.TestClassRunner.getDescription(TestClassRu
nner.java:57)
at
org.hansel.CoverageRunner.getDescription(CoverageRunner.java:119)
at
org.junit.internal.runners.CompositeRunner.getDescription(CompositeRu

This starts an infinite loop.

I'm using the latest JUnit 4 release, however I get the same results if
I use the JUnit distibution that is packaged with Hansel.

Any help would be appreciated.

Thanks,

Bob Butash (bob.butash@eds.com)

problem with .getSpecificationVersion()

2005-02-23T16:17:30Z

Hello,

in latest Hansel 1.02, I could not run my tests
(strange failures happened, from my TestCase's setUp()
methods, something like
java.lang.ExceptionInInitializerError. Found the reason
to be this code in a static initialize in one of my
classes:

Package pack = Package.getPackage("bla.blu.bli");
String vers = pack.getSpecificationVersion();

The second line of code did not work with Hansel ... No
idea if it's only preventing Hansel from working when
placed in a static initializer ...

Works great besides from my issue with this special
packages stuff!!! :-)

cu
Merten

Inner-inner classes not covered

2004-09-23T15:31:42Z

An inner-inner class is not instrumented. The reason is
the superfluous if (....$...) in ModifyingClassLoader.
A test case showing the problem is in the attachment:
The methods in the inner-inner classes should be shown
as not-covered.

... I'll repair this together with the categories, which I
hopefully will check in in a few days (after much more
testing).

Instruction: ifle[158](3) -> iconst_0 unknown

2004-09-23T15:16:06Z

See the attached file for an example which throws the
exception

java.lang.UnsupportedOperationException: Instruction:
ifle[158](3) -> iconst_0 unknown
at
org.hansel.stack.ComplexAnd.translateBranch
(ComplexAnd.java:138)
at org.hansel.stack.ComplexAnd.getStackEntry
(ComplexAnd.java:59)
at org.hansel.stack.ComplexOr.getStackEntry
(ComplexOr.java:44)
at
org.hansel.stack.StackFactory.checkComplexOperator
(StackFactory.java:180)
at
org.hansel.stack.StackFactory.calculateStackEntries
(StackFactory.java:203)
at
org.hansel.stack.StackFactory.calculateMethodStacks
(StackFactory.java:82)
at org.hansel.Instrumenter.instrument
(Instrumenter.java:291)
at org.hansel.Instrumenter.instrument
(Instrumenter.java:62)
at org.hansel.ModifyingClassLoader.modifyClass
(ModifyingClassLoader.java:205)
at org.apache.bcel.util.ClassLoader.loadClass
(ClassLoader.java:152)
at java.lang.ClassLoader.loadClass
(ClassLoader.java:255)
at org.hansel.ModifyingClassLoader.<init>
(ModifyingClassLoader.java:108)
at
org.hansel.ClassLoaderFactory$InternalClassLoader.<init>
(ClassLoaderFactory.java:113)
at
org.hansel.ClassLoaderFactory.createInternalClassLoader
(ClassLoaderFactory.java:57)
at
org.hansel.ClassLoaderFactory.createClassLoader
(ClassLoaderFactory.java:75)
at
org.hansel.CoverageDecorator.internalAddTest
(CoverageDecorator.java:103)
at org.hansel.CoverageDecorator.addTestClass
(CoverageDecorator.java:118)
at org.hansel.CoverageDecorator.<init>
(CoverageDecorator.java:66)
at org.hanseltest.TestInt.suite
(TestInt.java:31)
at org.hanseltest.AllTests.suite
(AllTests.java:98)
at
sun.reflect.NativeMethodAccessorImpl.invoke0(Native
Method)

Failed to invoke suite(): java.lang.NoSuchMethodError: junit

2004-09-22T11:31:17Z

Hi,

I have downloaded Hansel 1.0 and we are using JDK1.4.
I have created a class Example(which is to be tested)
and a TestCase class ExampleTest.

In the suite() method of the TestCase class I have
included the code for initializing the Hansel code
coverage class as follows:
public static Test suite() {

CoverageDecorator cd = new CoverageDecorator
(new Class[] { Example.class });

cd.addTestClass(ExampleTest.class);

cd.setDisplayStatistics(true);

return cd;

}

The exception that I get while running the ExampleTest
class as a JUnit class using WSAD is as follows:

Failed to invoke suite(): java.lang.NoSuchMethodError:
junit.framework.TestSuite: method <init>
(Ljava/lang/Class;Ljava/lang/String;)V not found

How can this problem be resolved?

Thanks and Regards,
Ranjana

Strange conditions for double compares, e.g. !(x < 0.0 != 0)

2004-09-17T07:53:59Z

For double comparions, hansel creates strange
conditions. For example,

if (x == 0) ...

where x is a double, creates the output

Coverage failure: Branch not completely covered.
Condition '!(x < 0.0 != 0)' is not fulfilled.
at org.hanseltest.TestDoubleClass.comparisons
(TestDoubleClass.java:40)

The reason is that the dcmpl and dcmpg instructions
work by themselves pushing 0, -1, or 1 on the stack,
which is then checked by an if, e.g.

0 dload_1
1 dconst_0
2 dcmpl
3 ifne 17

Here is a test which tests some more cases:

-------------

package org.hanseltest;

import junit.framework.TestCase;
import junit.framework.Test;

import org.hansel.CoverageDecorator;

/**
* Strange conditions for double comparisons
* @author HMMueller
*/
public class TestDouble extends TestCase {
/**
* Create a new Test.
* @param name Name of the test.
*/
public TestDouble(String name) {
super(name);
}

/** Cover the class. */
public void testSomething() {

new TestDoubleClass().comparisons(3);
}

/**
* Returns the coverage test suite.
* @return Test suite.
*/
public static Test suite() {
return new CoverageDecorator(TestDouble.class,
new Class[] {
TestDoubleClass.class });
}
}

class TestDoubleClass {
int comparisons(double x) {
double y = -x;
if (x == 0) {
return 15;
} else if (x < 0) {
return 16;
} else if (x <= 0) {
return 17;
} else if (x <= 2) {
return 18;
} else if (x <= -3) {
return 19;
} else if (x == -3) {
return 20;
} else if (y > 0) {
return 21;
} else if (y >= 0) {
return 22;
} else if (y >= -2) {
return 23;
} else {
return 99;
}
}
}

-------------

Regards
HMMueller

Should work on suite level, not just single testcase level

2003-11-15T21:47:55Z

If I want to refactor my TestCases and test fixtures
(setUp()) so that multiple TestCases work together to
provide complete coverage, Hansel doesn't support this.

Here is a post I made on the Test Driven Development
(http://groups.yahoo.com/group/testdrivendevelopment
/) newsgroup that discusses the problem:

------------------------

Hi,

There's a little bit of intro to this question and problem:

I've been reading Dave Astel's TDD book, and I find the
approach of test refactoring very interesting. He
advocates (paraphrased from memory) that you split up
your TestCases so that your fixtures apply to ALL the
tests in the TestCase. For example, TestList and
TestEmptyList, both with different setUp() methods for
the different fixtures.

I've also been using Hansel
(http://sourceforge.net/projects/hansel), which is a
code coverage tool. It ensures that a SINGLE given
TestCase executes every line and logic path in one or
more Classes Under Test. I've found this tool very
useful in helping me to TDD and drive towards simple,
well-designed tests and classes.

Here's the problem:

These two approaches seem to be incompatible. Hansel
only works on a single-testcase level, so if you split your
tests into two TestCase classes, they will both report
incomplete coverage, even though they may be working
together to provide complete coverage.

I know that Hansel could probably be modified to check
for coverage at a suite level, and I could then define
suites that group the TestCases that work together to
provide complete coverage. However, this probably
won't happen soon.

Does anyone have a solution to this problem? Does
anyone know of any FREE tools which provide the same
type of fine-grained, per-testcase coverage that Hansel
does, but also work at the suite level? NoUnit might
work, but I couldn't get it to work right the first time I
tried. Clover would probably be good too, but it's not
free.

ALso, the problem with NoUnit, Clover, and other
project-level coverage tools is that they lack the
immediate unit-level feedback that Hansel gives you. I
want to see immediately (after a single testcase run)
when I no longer have 100% coverage, not after
running the entire suite of tests. Plus, Hansel
integrates with my Eclipse junit to take me right to the
offending line when I click the teset failure.

Thanks,
Chad

Code seems to be uncoverable

2003-07-26T12:52:50Z

I get the following error message:

org.hansel.Probe$CoverageError: Coverage failure:
Branch not completely covered. Expression '!pieceToMove
instanceof scamsoft.games.spazchess.piece.Pawn' is
never false.
Coverage failure: Branch not completely covered.
Expression '!pieceToMove instanceof
scamsoft.games.spazchess.piece.Pawn' is never false.
at
scamsoft.games.spazchess.Board.determineMoveValidity(Board.java:595)
at
scamsoft.games.spazchess.Board.determineMoveValidity(Board.java:595)

But if I step through with a debugger, both options for
the branch in question are definitely executed.

The fact that it only reports 1 line in its stack
trace, but reports it twice, is strange for a start.

I have attached all the source code that should be
needed to compile and run the tests in
scamsoft.games.spazchess.test.SpazChessTestSuite.java.
It will need JUnit in the classpath.

Let me know if this is enough information to track the
problem down. It's possible that the error is all mine
as it started occurring quite late at night, so the
relevant bits of code to look at are all in the package
scamsoft.games.spazchess. They are Board.java:line 593,
and test.BoardTest.java, test.SpazChessTestSuite.java
and piece.test.*.java (these are all involved in
exercising the Board class directly or indirectly).

Thanks in advance!

P.S: Sorry the test code is such a copy-and-paste
nightmare. Now I look at it, I have no idea why I did
that, and it will be fixed soon, trust me! :-)

".class" causes coverage failure (e. g. "Integer.class")

2003-07-23T12:29:57Z

Similar to alreay reported bugs I get a coverage failure,
if I use the following code in any of my coverage test
classes (as attribute or within a method):

...
Class c = Integer.class;
...

I am using JBuilder 8 eval version with jdk 1.4.1 and
hansel 1.0 (released version from 06/15/2003).

If I replace the code by calling a factory method of
another class that is not coverage tested, there is no
coverage failure reported by hansel:

...
Class c = ClassFactory.getIntegerClass();
...

Can anybody reproduce this problem?

>= creates wrong expression text

2003-06-21T07:05:23Z

Testing the following

public void cmp(int a, int b)
{
if (a < b)
System.out.println("b >= a is " + (b >= a));
}

with

cmp(1, 2)

yields the following output:

-----------------
.b >= a is true
....F.F
Time: 0,02
There were 2 failures:
1) Probe in "TryHansel.java" line 128[...] 'b >= a' is
never true.
at TryHansel.cmp(TryHansel.java:128)
2) Probe in "TryHansel.java" line 129[...] 'a < b' is never
true.
at TryHansel.cmp(TryHansel.java:129)

FAILURES!!!
Tests run: 6, Failures: 2, Errors: 0
-----------------

The important lines are the first and fifth:

.b >= a is true
[...]
1) [...] 'b >= a' is never true.

which is a contradiction. It seems that the expression
creation for >= (and also >, <=, < ??) swaps the left
and right side. The output

1) ... 'a >= b' is never true.

would be correct.