On 11/12/2015 07:14 PM, Jörg Schaible wrote:
> Hi Thomas,
> 
> Thomas Neidhart wrote:
> 
>> Hi all,
>>
>> in order to provide a work-around for the known remote code exploit via
>> java de-serialization of malicious InvokerTransformer instances, I would
>> like to start a vote to release Commons Collections 3.2.2 based on RC2.
>>
>> Notes:
>>
>>  * the site will not be published, it just serves as a reference to
>> access the various reports. After a successful vote, the current 4.X
>> branch site will be updated with relevant information and published.
>>
>>  * some tests might fail with various IBM JDK 6 JREs, these are known
>> issues and have been worked-around in the 4.X branch but are not
>> back-ported to this release.
>>
>>  * Collections 3.2.2 can not be compiled with JDK 8 due to a name clash
>> with a newly introduced default method in the Map interface.
>>
>>  * the collections-testframework.jar that has been published in previous
>> versions is not included in this release
>>
>>
>> Changes from RC1:
>>
>>  * fixed RAT report
>>  * fixed NOTICE file
>>  * improve the security fix: it has been made symmetric in the sense
>>    that also the serialization of an unsafe class is disabled by
>>    default and will result in an exception
>>  * changed the system property to re-enable serialization of unsafe
>>    classes. It is now
>>    "org.apache.commons.collections.enableUnsafeSerialization"
>>  * all classes in the functor package which (based on current
>>    knowledge) have to be considered unsafe cannot be serialized/
>>    de-serialized any more by default. This includes the following
>>    classes:
>>
>>  ** CloneTransformer
>>  ** PrototypeFactory (inner classes
>>                       PrototypeCloneFactory and
>>                       PrototypeSerializationFactory)
>>  ** InstantiateFactory
>>  ** InstantiateTransformer
>>  ** ForClosure
>>  ** WhileClosure
>>  ** InvokerTransformer
>>
>>
>>
>> Collections 3.2.2 RC2 is available for review here:
>>     https://dist.apache.org/repos/dist/dev/commons/collections/
>>     (svn revision 11147)
>>
>> Maven artifacts are here:
>>
>>
> https://repository.apache.org/content/repositories/orgapachecommons-1116/commons-collections/commons-collections/3.2.2/
>>
>> Details of changes since 3.2.1 are in the release notes:
>>
>> https://dist.apache.org/repos/dist/dev/commons/collections/RELEASE-NOTES.txt
>>
>> http://people.apache.org/builds/commons/collections/3.2.2/RC2/changes-report.html
>>
>> The tag is here:
>>
>>
> https://svn.apache.org/repos/asf/commons/proper/collections/tags/COLLECTIONS_3_2_2_RC2
>>     (svn revision 1713883)
>>
>> Site:
>>     http://people.apache.org/builds/commons/collections/3.2.2/RC2/
>>
>> Clirr Report (compared to 3.2.1):
>>
>> http://people.apache.org/builds/commons/collections/3.2.2/RC2/clirr-report.html
>>
>> RAT Report:
>>
>> http://people.apache.org/builds/commons/collections/3.2.2/RC2/rat-report.html
>>
>> KEYS:
>>   https://www.apache.org/dist/commons/KEYS
>>
>> Please review the release candidate and vote.
>>
>>
>> Considering that this is a security related release and that RC1 did not
>> show any functional problems with the release, I plan to close this vote
>> in 24 from now, i.e. after 1800 GMT 12-November 2015
>>
>>   [ ] +1 Release these artifacts
>>   [ ] +0 OK, but...
>>   [ ] -0 OK, but really should fix...
>>   [ ] -1 I oppose this release because...
> 
> -1,
> 
> sorry, but there's a regression
> 
> The package claims to be compatible with Java 1.3. Well, I don't have 1.3 
> anymore, but 1.4. And I can build CC-3.2.1 and run all tests with Blackdown 
> JDK 1.4 and Maven 2.0.11.
> 
> For CC-3.2.2 I have to use at least Java 5 and Maven 3.0(.5):
> 
> - Using java-1.4 profile: Build fails, because tests no longer compile
> - Sun JDK 1.5: TestAllPackages fails due to SecurityException:
> ================== %< ==================
> Running org.apache.commons.collections.TestAllPackages
> java.lang.SecurityException
>         at 
> org.apache.commons.collections.TestExtendedProperties$1.checkPropertyAccess(TestExtendedProperties.java:322)
>         at java.lang.System.getProperty(System.java:628)
>         at 
> sun.security.action.GetPropertyAction.run(GetPropertyAction.java:66)
>         at java.security.AccessController.doPrivileged(Native Method)
>         at java.io.PrintWriter.<init>(PrintWriter.java:77)
>         at java.io.PrintWriter.<init>(PrintWriter.java:61)
>         at 
> org.apache.maven.surefire.report.LegacyPojoStackTraceWriter.writeTraceToString(LegacyPojoStackTraceWriter.java:56)
>         at 
> org.apache.maven.surefire.booter.ForkingRunListener.encode(ForkingRunListener.java:330)
>         at 
> org.apache.maven.surefire.booter.ForkedBooter.main(ForkedBooter.java:119)
> ================== %< ==================
> - Sun JDK 1.6: OK
> - Oracle JDK 1.7: OK
> - IBM JDK 1.5: OK (!!)
> - IBM JDK 1.6 (J9 2.4): fails (as expected, same for CC-3.2.1)
> - IBM JDK 1.7: OK (!!)
> - IcedTea 6 (OpenJDK): TestAllPackages fails due to SecurityException:
> ================== %< ==================
> Running org.apache.commons.collections.TestAllPackages
> java.lang.SecurityException
>         at 
> org.apache.commons.collections.TestExtendedProperties$1.checkPropertyAccess(TestExtendedProperties.java:322)
>         at java.lang.System.getProperty(System.java:628)
>         at 
> sun.security.action.GetPropertyAction.run(GetPropertyAction.java:66)
>         at java.security.AccessController.doPrivileged(Native Method)
>         at java.io.PrintWriter.<init>(PrintWriter.java:77)
>         at java.io.PrintWriter.<init>(PrintWriter.java:61)
>         at 
> org.apache.maven.surefire.report.LegacyPojoStackTraceWriter.writeTraceToString(LegacyPojoStackTraceWriter.java:56)
>         at 
> org.apache.maven.surefire.booter.ForkingRunListener.encode(ForkingRunListener.java:330)
>         at 
> org.apache.maven.surefire.booter.ForkedBooter.main(ForkedBooter.java:119)
> ================== %< ==================
> - IcedTea 7 (OpenJDK): OK
> 
> 
> TestExtendedProperties.testActiveSecurityManager is the only test using a 
> SM, but I wonder, why it fails the test now, because both failing JDKs have 
> no problem building CC-3.2.1 (using Maven 3.0.5) and all tests pass fine.

ok, the errors have been fixed in the branch.

I have successfully tested it with the Oracle/Sun jdk from 1.4 till 1.7

The Jdk 1.3 does not run anymore on my computer.

Maybe you have the time to execute the tests again from trunk, I will
create a new RC in about 2 hours.

Thomas

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
For additional commands, e-mail: dev-h...@commons.apache.org

Reply via email to