As a minor clean up facilitating debugging, I've refactored the unit test fork attribute in a default.properties property named "test.fork", the default is true, as in the build.xml file.
Gary -----Original Message----- From: Gary Gregory [mailto:[EMAIL PROTECTED] Sent: Thursday, May 26, 2005 5:37 PM To: Steven Caswell Cc: Jakarta Commons Developers List Subject: RE: FW: [lang] DateUtils test fails under 1.2.2 and not under 1.4.2 Hi: I found possible solution. With 1.3.1, if I set fork to false, the test is found and run. Weird. Gary ________________________________ From: Steven Caswell [mailto:[EMAIL PROTECTED] Sent: Thursday, May 26, 2005 5:06 PM To: Gary Gregory Cc: Jakarta Commons Developers List Subject: Re: FW: [lang] DateUtils test fails under 1.2.2 and not under 1.4.2 That is so very odd. I'm at a loss to explain it. I don't have access to a system with JDK 1.3 so I'm not able to try it. Maybe someone else can try? One thing I did find I needed to do was to change th junit version in default.properties from 3.7 to 3.8.1. Gary, could you try making this change and building under 1.3.1? I'm just curious to see if this makes a difference, though I don't see why it should. Thanks. On 5/25/05, Gary Gregory <[EMAIL PROTECTED]> wrote: Steven: I updated from CVS today and got new build.xml. I still get the error below on Sun Java 1.3.1_14. Thanks, Gary ________________________________ From: Gary Gregory Sent: Tuesday, May 24, 2005 7:12 PM To: 'Steven Caswell' Cc: Jakarta Commons Developers List Subject: RE: [lang] DateUtils test fails under 1.2.2 and not under 1.4.2 Steven: (You're welcome.) Downloaded RC6a-src and I am now experiencing Weirdness: I run "ant clean dist-build-2.1 test" which works fine on Sun Java 1.4.2_08. On Sun Java 1.3.1_14 I get: test.lang: [echo] Running lang package tests ... [java] Class not found "org.apache.commons.lang.LangTestSuite" BUILD FAILED C:\temp\commons-lang-2.1-RC6\build.xml:166: Java returned: 1 Which makes no sense at first glance. Gary ________________________________ From: Steven Caswell [mailto:[EMAIL PROTECTED] Sent: Tuesday, May 24, 2005 7:00 PM To: Gary Gregory Cc: Jakarta Commons Developers List Subject: Re: [lang] DateUtils test fails under 1.2.2 and not under 1.4.2 Gary, I put up the revised source distribution in http://www.apache.org/~stevencaswell/commons-lang-2.1 <http://www.apache.org/%7Estevencaswell/commons-lang-2.1> I named them commons-lang-2.1-RC6a-src.* so you could easily tell they are modified. I was just too lazy to change it to RC7 for a minor non-code fix, and since there will probably be an RC7 with the date utils test change. Thanks for your patience and thanks for testing these things. On 5/24/05, Steven Caswell <[EMAIL PROTECTED] > wrote: Gary, I have a fix for that problem. I'll go ahead and put up a corrected source distribution with the fix and without the DateUtilsTest correction so you can try 1.3.1. On 5/24/05, Gary Gregory <[EMAIL PROTECTED] > wrote: I'd like to check 1.3.1 but... running "ant clean build-dist-2.1 test" from src-zip does not make it past text tests: test.text: [echo] Running text package tests ... [java] Class not found " org.apache.commons.lang.text.TextTestSuite" BUILD FAILED C:\temp\commons-lang-2.1-RC6\build.xml:206: Java returned: 1 Which should not be run in the 1st place... (Ant 1.6.4 and Java 1.4.2_08) Gary -----Original Message----- From: Steven Caswell [mailto:[EMAIL PROTECTED] Sent: Tuesday, May 24, 2005 5:39 PM To: Jakarta Commons Developers List Subject: [lang] DateUtils test fails under 1.2.2 and not under 1.4.2 All, I have discovered a group of tests in the DateUtils test case that fails under 1.2.2 but not under 1.4.2. They are testing the DateUtils round method when rounding a date in the MET timezone across the start and end of DST. If I remember correctly, this was in response to a user bug report. In particular the test is at line 472 of the testRound method in DataUtilsTest.java. The test passes fine when run under 1.4.2 but fails under 1.2.2. I don't have a 1.3 installation available so I don't know if it passes or fails under 1.3. I propose that we put a condition around the test so that is only run when the Java version is 1.4, and add a note to the round method javadoc and the release notes stating that the round method may not work properly in all cases involving DST rollovers in previous JVMs, with this case as an example. Thoughs? -- Steven Caswell [EMAIL PROTECTED] Take back the web - http://www.mozilla.org -- Steven Caswell [EMAIL PROTECTED] Take back the web - http://www.mozilla.org -- Steven Caswell [EMAIL PROTECTED] Take back the web - http://www.mozilla.org -- Steven Caswell [EMAIL PROTECTED] Take back the web - http://www.mozilla.org --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
