Apache MRUnit 1.0.0 发布了,包含大量改进和 bug 修复,详情请看发行说明。
MRUnit是由Couldera公司开发的专门针对 Hadoop中 编写MapReduce单元测试的框架,基本原理是JUnit4和 EasyMock。MR就是Map和Reduce的缩写。MRUnit框架非常精简,其核心的单元测试依赖于JUnit。而且MRUnit实现了一套 Mock对象来控制OutputCollector的操作,从而可以拦截OutputCollector的输出,和我们的期望结果进行比较,达到自动断言 的目的。
Sub-task
[MRUNIT-144] - Change repository location on MRunit website
[MRUNIT-145] - Change repository location on Jenkins
[MRUNIT-146] - Change repository location in source code
[MRUNIT-172] - Branch trunk should use hadoop 2
[MRUNIT-173] - Create instructions on building hadoop1 and hadoop2 source/binary tarball
[MRUNIT-174] - Remove portions of MRUNIT-161 which don't compile on trunk-hadoop1
[MRUNIT-176] - Branch trunk-hadoop1 should use hadoop 1
[MRUNIT-183] - Binary distribution should include only jar with correct classifier
Bug
[MRUNIT-105] - testJavaSerialization methods fail in Java 7 because they do not set a group comparator
[MRUNIT-114] - should support calling configure on sorting and grouping comparators if applicable
[MRUNIT-115] - set JobConf specific to the inputformat reading the results of the outputformat/copying when not using a real outputformat
[MRUNIT-116] - Setting of configuration on Mappers/Reducers in MapDriver and ReduceDriver is inconsistent
[MRUNIT-124] - DistributedCache should move files from getCacheFiles to getLocalCacheFiles
[MRUNIT-126] - Counter tests are ignored if test is executed using run() rather than runTest()
[MRUNIT-127] - Key grouping with GroupingComparators is not consistent with MapReduce behaviour
[MRUNIT-130] - Download page does not point to KEYS; no hashes in download area; must not advertise snapshot builds
[MRUNIT-131] - Comparators registered using WritableComparator.define are not used during tests
[MRUNIT-132] - Spelling mistakes in TestReducerInputValueReuse
[MRUNIT-159] - MRUnit not compiling with hadoop 0.23.3
[MRUNIT-161] - some distributed cache apis not working - JobContext getCacheArchives()
[MRUNIT-163] - getTaskAttemptID is missing from context
[MRUNIT-165] - MapReduceDriver calls Mapper#cleanup for each input instead of once
[MRUNIT-167] - context passed to an OuputFormatter is ignored by the framework
[MRUNIT-178] - Remove portions of MRUNIT-167 on trunk-hadoop1 which do not compile
[MRUNIT-179] - Set mapred.output.dir and mapred.input.dir in trunk
思朴互联-西部数码代理站
2013-4-16