[ORC-410] Fix a locale-dependent test in TestCsvReader

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.5.4, 1.6.0
• Assignee: Kotaro Terada
• Reporter: Kotaro Terada
• Created: 28/Sep/18 07:15
• Updated: 02/Jan/19 19:12
• Resolved: 01/Oct/18 16:29

Description

testCustomTimestampFormat in TestCsvReader fails in some environments because the test is locale-dependent.

In this test, we try to parse a DateTime string (such as '21 Mar 2018 12:23:34') with a given timestamp format. The problem is that English month abbreviations (such as 'Mar') are locale-dependent. When the locale of Java Virtual Machine is a locale where the language is English (e.g., en_US and en_GB), this test passes without any problems. However, when the locale of JVM is a locale where the language is non-English (e.g., ja_JP and zh_CN), the test fails as follows.

[INFO] Running org.apache.orc.tools.convert.TestCsvReader
[ERROR] Tests run: 5, Failures: 0, Errors: 1, Skipped: 0, Time elapsed: 0.237 s <<< FAILURE! - in org.apache.orc.tools.convert.TestCsvReader
[ERROR] testCustomTimestampFormat(org.apache.orc.tools.convert.TestCsvReader)  Time elapsed: 0.143 s  <<< ERROR!
org.threeten.bp.format.DateTimeParseException: Text '21 Mar 2018 12:23:34' could not be parsed at index 3
        at org.apache.orc.tools.convert.TestCsvReader.testCustomTimestampFormat(TestCsvReader.java:189)

Issue Links

• links to: GitHub Pull Request #314