> On Nov. 5, 2013, 7:31 p.m., Brock Noland wrote: > > This looks very good! Why didn't we add a new test to TestHiveDecimal? It > > looks like this functionality could be tested there. > > Xuefu Zhang wrote: > Were you referring to the change made to HiveDecimal? If so, yes, that > can be tested in TestHiveDecimal. However, the main objection of the the > patch cannot be tested in TestHiveDecimal, as HiveDecimal represents data > only, but type info is about metadata.
Sorry I wasn't clear. Yes I am asking why we didn't add new test functionality to TestHiveDecimal for the changes in HiveDecimal. The change looks like it's a good candidate for a unit test to ensure we don't regress. - Brock ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/15217/#review28202 ----------------------------------------------------------- On Nov. 4, 2013, 11:40 p.m., Xuefu Zhang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/15217/ > ----------------------------------------------------------- > > (Updated Nov. 4, 2013, 11:40 p.m.) > > > Review request for hive. > > > Bugs: HIVE-5726 > https://issues.apache.org/jira/browse/HIVE-5726 > > > Repository: hive-git > > > Description > ------- > > With this patch, the resuting schema from constant is in accordance with the > constant instead of the default from the type. > > > Diffs > ----- > > common/src/java/org/apache/hadoop/hive/common/type/HiveDecimal.java 606f6d4 > ql/src/java/org/apache/hadoop/hive/ql/parse/TypeCheckProcFactory.java > 47d7995 > ql/src/test/queries/clientpositive/ctas_hadoop20.q 4961b97 > ql/src/test/results/clientpositive/ctas_hadoop20.q.out 885e985 > ql/src/test/results/clientpositive/literal_decimal.q.out 2d1ae55 > > Diff: https://reviews.apache.org/r/15217/diff/ > > > Testing > ------- > > New tests are added. Test suite passed. > > > Thanks, > > Xuefu Zhang > >
