|
[
Permalink
| « Hide
]
Myrna van Lunteren added a comment - 04/Aug/08 08:12 PM
not sure exactly when this started failing; the nightly tests haven't been finishing before the next cycle kicks off (last good result: http://people.apache.org/~myrnavl/derby_test_results/main/windows/testSummary-678612.html.)
I must've been running with sane jars when I saw this.
There's been good nightly results since, but the test hasn't failed, because it gets skipped with insane jars. (See The test still fails. Actually, I can make the same failure happen with sane jars and sun's jdk142.
I think it's because of this characteristic of PrintWriter.println: (from the api doc:) println() methods use the platform's own notion of line separator rather than the newline character println is used in this section in org.apache.derby.shared.common.sanity.AssertFailure: } catch (NoSuchMethodException e) { p.println("(Skipping thread dump because it is not " + "supported on JVM 1.4)"); And then the comparison in AssertFailureTest fails on the windows platform because the line separator is different from what's given as expected: String expected = "(Skipping thread dump because it is not " + "supported on JVM 1.4)\n"; Using \r\n in the expected string makes the assertEquals pass on windows. A similar check for correct string returned passes for 1.5 and higher because the comparison there is using s.startsWith(). The easiest solution is to trim() s before doing the comparison. committed the change with revision 793848.
backported the fix to 10.5.2.1 with revision 800646.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||