Re: [Springframework-developer] Re: Eclipse problem when running entire test suite

[Colin S. <col...@ex...>]

Command-line, or in Eclipse?  It doesn't fail for me in either. The only 
tests which fail for me right now, in both environments, are 
CompilerTests and AttributeWriterTests.



jürgen höller [werk3AT] wrote:

>BTW, I just updated from CVS, and ProxyFactoryTests fails now...
>
>
>-----Original Message-----
>From: spr...@li...
>[mailto:spr...@li...]On Behalf
>Of Rod Johnson
>Sent: Tuesday, November 18, 2003 3:48 PM
>To: spr...@li...
>Subject: Re: [Springframework-developer] Re: Eclipse problem when
>running entire test suite
>
>
>Colin,
>
>Thanks very much for tracking this problem down and fixing it.
>
>I've fixed the single test that was failing, which relied on a static that
>was not reset before that test. So now all tests pass in Eclipse again, and
>I can run the whole test suite in 30 seconds vs 3 minutes from Ant. Trivial
>to run _all_ tests instead of just a few before any commit.
>
>Regards,
>Rod
>
>----- Original Message ----- 
>From: "Colin Sampaleanu" <col...@ex...>
>To: <spr...@li...>
>Sent: Tuesday, November 18, 2003 3:38 AM
>Subject: [Springframework-developer] Re: Eclipse problem when running entire
>test suite
>
>
>  
>
>>There was only one test actually calling System.setProperties, and
>>wiping out the System properties. I made that test restore them
>>afterwards, and now when running the entire suite in Eclipse all the
>>Errors are gone. So at this point there is only one test which fails in
>>Eclipse and not via command-line, and only when running the entire test
>>suite. That's
>>  ListableBeanFactoryImplTestSuite
>>I presume that for this as well there is interference from another test,
>>but I don't have time to look at it.
>>
>>
>>
>>Colin Sampaleanu wrote:
>>
>>    
>>
>>>I had a few minutes now to take a look at this. I figured out what is
>>>hosing the NestedCheckedExceptionTests (possibly the others too, but I
>>>just looked at the first one).
>>>
>>>When run as part of the test suite, it dies with a
>>>NullPointerException on this line:
>>>            nce.printStackTrace(pw);
>>>and it dies because it eventually gets and tries to use the System
>>>property 'line.separator'. The problem is that one of the other tests
>>>has killed all the normal system properties.
>>>
>>>Looks like Eclipse runs all tests in the same 'run' in the same vm (as
>>>opposed to our ant build script, which calls 'batchtest' with
>>>fork="yes").
>>>
>>>Now one solution would be to make sure all tests restore system
>>>properties after overriding them. Alternately, does anybody know how
>>>to force Eclipse to use a new vm for each test in a testsuite? I am
>>>using 3.0M4, and there doesn't seem to be any option for this.
>>>
>>>
>>>Colin Sampaleanu wrote:
>>>
>>>      
>>>
>>>>If I do a 'rebuild all', then run all tests, I get the following
>>>>failures:
>>>>Fails:
>>>>
>>>>        
>>>>
>testFactoryBeanDidNotCreatePrototype(org.springframework.beans.factory.Lista
>bleBeanFactoryImplTestSuite)
>  
>
>>>>testAttributes(org.springframework.metadata.bcel.AttributeWriterTests)
>>>>
>>>>        
>>>>
>testAbbreviatedAttribute(org.springframework.metadata.bcel.AttributeWriterTe
>sts)
>  
>
>>>>testMethods(org.springframework.metadata.bcel.AttributeWriterTests)
>>>>testFields(org.springframework.metadata.bcel.AttributeWriterTests)
>>>>testSingleFile(org.springframework.metadata.qdox.CompilerTests)
>>>>Errors:
>>>>
>>>>        
>>>>
>testWithApplicationContext(org.springframework.web.multipart.commons.Commons
>MultipartResolverTests)
>  
>
>>>>        
>>>>
>testWithServletContextAndFilter(org.springframework.web.multipart.commons.Co
>mmonsMultipartResolverTests)
>  
>
>>>>        
>>>>
>testWithServletContextAndFilterWithCustomBeanName(org.springframework.web.mu
>ltipart.commons.CommonsMultipartResolverTests)
>  
>
>>>>testNoRootCause(org.springframework.core.NestedCheckedExceptionTests)
>>>>testRootCause(org.springframework.core.NestedCheckedExceptionTests)
>>>>
>>>>Now AttributeWriterTests and CompilerTests fail from command-line ant
>>>>already. The other 3 test cases (6 tests total) all execute ok if run
>>>>individually, but fail if run as part of running the whole suite.
>>>>Very weird. What is also annoying is that on the errors, Eclipse
>>>>doesn't actually give any details as to what the problem is when
>>>>running multiple tests. Rod, I don't even know how you figured out
>>>>it's a NoClassDefFoundError. The only time I've ever been able to
>>>>print out an error in that case was when running one test case
>>>>individually, but in this case they all work fine individually.
>>>>
>>>>
>>>>
>>>>Rod Johnson wrote:
>>>>
>>>>        
>>>>
>>>>>It seems to be a NoClassDefFoundError, and it affects mainly the
>>>>>Hibernate
>>>>>call to config.buildSessionFactory() and anything to do with CGLIB.
>>>>>
>>>>>I don't have time to dig deeper right now, but does anyone have any
>>>>>tips?
>>>>>
>>>>>Regards,
>>>>>Rod
>>>>>          
>>>>>