Hi Joerg, the failure for JDK 1.8 is clear: this version updated the Map interface with a new method boolean remove(Object, Object) which clashes with the ones in MultiValueMap and MultiMap. We should change the return type to be compatible.
For IBM JDK 6: I did already ignore several tests (due to problems with the TreeMap implementation there) if run with this JVM but obviously something has changed since then, need to check. For IBM JDK 5: It only fails to compile the tests, but I could not yet figure out why. Thomas On Wed, Nov 13, 2013 at 12:13 AM, Jörg Schaible <joerg.schai...@gmx.de>wrote: > Hi Thomas, > > I've downloaded the source tar.gz and tested it with my compiler zoo. > > 1/ IBM JDK 1.5 fails with compilation error (without telling where or why). > 2/ Sun JDK 1.5 compiles and runs tests fine, using these class files to run > the tests: > 2.1/ IBM JDK 1.6 runs tests with 5 failures / 15 errors > 2.2/ IBM JDK 1.7 runs tests successful > 2.3/ IcedTea 6 / OpenJDK runs tests successful > 2.4/ IcedTea 7 / OpenJDK runs tests successful > 2.5/ Oracle JDK 1.7 runs tests successful > 2.6/ Oracle JDK 1.8 (b109) runs tests successful > 2.7/ Sun JDK 1.6 runs tests successful > 3/ IBM JDK 1.6 compiles and runs tests with 5 failures / 15 errors > 4/ IBM JDK 1.7 compiles and runs tests successful > 5/ IcedTea 6 compiles and runs tests successful > 6/ IcedTea 7 compiles and runs tests successful > 7/ Sun JDK 1.6 compiles and runs tests successful > 8/ Oracle JDK 1.7 compiles and runs tests successful > 9/ Oracle JDK 1.8 (b109) fails to compile: > ======================= %< ========================= > [ERROR] Failed to execute goal org.apache.maven.plugins:maven-compiler- > plugin:3.1:compile (default-compile) on project commons-collections4: > Compilation failure: Compilation failure: > [ERROR] /home/joehni/tmp/download/commons-collections4-4.0- > src/src/main/java/org/apache/commons/collections4/map/MultiValueMap.java: > [211,14] remove(java.lang.Object,java.lang.Object) in > org.apache.commons.collections4.map.MultiValueMap cannot implement > remove(java.lang.Object,java.lang.Object) in java.util.Map > [ERROR] return type V is not compatible with boolean > [ERROR] /home/joehni/tmp/download/commons-collections4-4.0- > src/src/main/java/org/apache/commons/collections4/MultiMap.java:[64,14] > name > clash: remove(K,V) in org.apache.commons.collections4.MultiMap and > remove(java.lang.Object,java.lang.Object) in java.util.Map have the same > erasure, yet neither overrides the other > [ERROR] /home/joehni/tmp/download/commons-collections4-4.0- > src/src/main/java/org/apache/commons/collections4/map/MultiKeyMap.java: > [199,14] remove(java.lang.Object,java.lang.Object) in > org.apache.commons.collections4.map.MultiKeyMap cannot implement > remove(java.lang.Object,java.lang.Object) in java.util.Map > [ERROR] return type V is not compatible with boolean > ======================= %< ========================= > > Actually I don't know, if the older IBM compilers simply have problems with > generics or if J8 is representative here - at least it runs. > > Cheers, > Jörg > > Thomas Neidhart wrote: > > > Hi, > > > > I'd like to call a vote for releasing Commons Collections 4.0 based on > > RC3. > > > > Changes since RC2: > > > > * [COLLECTIONS-499] Refactored the test framework for Bag > > implementations to extend from "AbstractCollectionTest" by > > decorating the concrete Bag instance with a CollectionBag or > > CollectionSortedBag. > > * [COLLECTIONS-498] "CollectionBag" will now also respect the > > contract of the decorated bag in case a null argument is provided > > to either removeAll or retainAll. > > * [COLLECTIONS-497] Added bag decorator "CollectionSortedBag" which > > decorates a SortedBag to make it comply with the Collection > > contract. > > * [COLLECTIONS-496] "UnmodifiableBoundedCollection" does now also > > implement the marker interface "Unmodifiable" similar as all other > > unmodifiable decorators. > > * [COLLECTIONS-495] "UnmodifiableTrie#unmodifiableTrie(Trie)" will > > not decorate again an already unmodifiable Trie. Also the return > > type has been changed to "Trie" to be consistent with other > > Unmodifiable decorators. > > * [COLLECTIONS-494] Moved "Equator" interface to base package for > > consistency. Thanks to Emmanuel Bourg. > > * improved release notes with suggestions from Emmanuel Bourg > > * added unit tests for TrieUtils and QueueUtils and factory methods > > for all Unmodifiable decorators > > * added more unit tests for the Queue interface > > * cleanup of many test classes: removed unused imports, unneeded > > constructors > > > > Changes since RC1: > > > > * release notes are now also included in the binary distribution > > * removed spurious hashCode() method in AbstractPatriciaTrie, the > > inherited one from AbstractMap will be used instead > > > > > > Collections 4.0 RC3 is available for review here: > > https://dist.apache.org/repos/dist/dev/commons/collections/ > > (svn revision 3477) > > > > Maven artifacts are here: > > > > > > https://repository.apache.org/content/repositories/orgapachecommons-118/org/apache/commons/commons-collections4/4.0/ > > > > The tag is here: > > > > > > https://svn.apache.org/repos/asf/commons/proper/collections/tags/COLLECTIONS_4_0_RC3/ > > (svn revision 1540843) > > > > Site: > > http://people.apache.org/builds/commons/collections/4.0/RC3/ > > > > Details of changes can be found in the release notes: > > > > > https://dist.apache.org/repos/dist/dev/commons/collections/RELEASE-NOTES.txt > > > > > http://people.apache.org/builds/commons/collections/4.0/RC3/changes-report.html > > > > Please review the release candidate and vote. > > This vote will close no sooner than 72 hours from now. > > > > [ ] +1 Release these artifacts > > [ ] +0 OK, but... > > [ ] -0 OK, but really should fix... > > [ ] -1 I oppose this release because... > > > > Note: the clirr report is generated against 4.0-alpha1 and it is also > > highlighted in the release notes that this release is not compatible > > with any previous release. > > > > Note2: there have been reports in the past that some unit tests fail > > with certain versions of the IBM Java 6 VM. Some tests have been > > disabled when run with a IBM Java 6 VM due to some bugs in the > > java.util.TreeMap implementation, but it may still fail for other > > versions. > > > > Thank you for your reviews, > > > > Thomas > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > >
