castor

hashCode() and equals() infinitely recurse with cyclid IDREFs

Details

  • Type: Bug Bug
  • Status: Closed Closed
  • Priority: Major Major
  • Resolution: Fixed
  • Affects Version/s: 1.1.1
  • Fix Version/s: 1.2
  • Component/s: XML code generator
  • Labels:
    None
  • Environment:
    WIndows XP SP2/Linux/OSX j1.4,1.5,1.6
  • Testcase included:
    yes
  • Number of attachments :
    4

Description

Elements with elements or attributes of type IDREF or IDREFs that for a particular instance resolve to any parent element will result in an StackOverflowError when either the hashCode or equals methods are called.

  1. Zip Archive
    bug2012.zip
    11/Jun/07 3:34 PM
    4 kB
    Jim Procter
  2. Java Source File
    CycleBreaker.java
    13/Jun/07 9:41 AM
    3 kB
    Jim Procter
  3. Text File
    patch-C2012-20070804.txt
    04/Aug/07 9:45 AM
    10 kB
    Ralf Joachim
  4. Text File
    SourceFactorybug2012alltests.patch
    13/Jun/07 9:35 AM
    5 kB
    Jim Procter

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Jim Procter added a comment -

test case exhibiting cyclic IDREF in hashCode and equals methods

Show
Jim Procter added a comment - test case exhibiting cyclic IDREF in hashCode and equals methods
Hide
Permalink
Jim Procter added a comment -

patch for SourceFactory version from SVN tagged with castor-1.1

Show
Jim Procter added a comment - patch for SourceFactory version from SVN tagged with castor-1.1
Hide
Permalink
Jim Procter added a comment -

org.castor.util.CycleBreaker class implementing the thread safe cycle detection methods used in the SourceGenerator patch.

Show
Jim Procter added a comment - org.castor.util.CycleBreaker class implementing the thread safe cycle detection methods used in the SourceGenerator patch.
Hide
Permalink
Jim Procter added a comment -

The attached patch and org.castor.util.CycleBreaker class are sufficient for the test case but cause one basic CTF test to fail:
...
Time: 136.282
There was 1 failure:
1) sourcegenerator/Primitives/WithWrappers/Test when using wrapper objects instead of primitives#Test Marshaling with the generated Descriptors_ReferenceDocument(org.castor.xmlctf.TestWithReferenceDocument)junit.framework.AssertionFailedError: The initial reference object and the one resulting of the marshal/unmarshal process are different
at org.castor.xmlctf.TestWithReferenceDocument.runTest(TestWithReferenceDocument.java:257)
at org.castor.xmlctf.CastorTestSuiteRunner.main(CastorTestSuiteRunner.java:167)

FAILURES!!!
Tests run: 304, Failures: 1, Errors: 0

I've not yet worked out quite where this is failing, but it is almost certainly due to the patched equals method behaving incorrectly. Hopefully someone more experienced than myself will spot the problem here.

Show
Jim Procter added a comment - The attached patch and org.castor.util.CycleBreaker class are sufficient for the test case but cause one basic CTF test to fail: ... Time: 136.282 There was 1 failure: 1) sourcegenerator/Primitives/WithWrappers/Test when using wrapper objects instead of primitives#Test Marshaling with the generated Descriptors_ReferenceDocument(org.castor.xmlctf.TestWithReferenceDocument)junit.framework.AssertionFailedError: The initial reference object and the one resulting of the marshal/unmarshal process are different at org.castor.xmlctf.TestWithReferenceDocument.runTest(TestWithReferenceDocument.java:257) at org.castor.xmlctf.CastorTestSuiteRunner.main(CastorTestSuiteRunner.java:167) FAILURES!!! Tests run: 304, Failures: 1, Errors: 0 I've not yet worked out quite where this is failing, but it is almost certainly due to the patched equals method behaving incorrectly. Hopefully someone more experienced than myself will spot the problem here.
Hide
Permalink
Jim Procter added a comment -

updated patch to SourceFactory which also passes all CTF tests.

Show
Jim Procter added a comment - updated patch to SourceFactory which also passes all CTF tests.
Hide
Permalink
Jim Procter added a comment -

additional usage javadoc relating to the outstanding CTF failure in the previous version of the SourceFactory patch where cycle breaking was incorrectly applied to identical references to a single string constant shared by the objects being compared.

Show
Jim Procter added a comment - additional usage javadoc relating to the outstanding CTF failure in the previous version of the SourceFactory patch where cycle breaking was incorrectly applied to identical references to a single string constant shared by the objects being compared.
Hide
Permalink
Ralf Joachim added a comment -

Final patch.

Show
Ralf Joachim added a comment - Final patch.

People

  • Assignee:
    Ralf Joachim
    Reporter:
    Jim Procter
Vote (0)
Watch (0)

Dates

  • Created:
    Updated:
    Resolved: