Did 'cvs update', 'bootstrap', and 'build test' again this morning, on my
work machine this time (Win2K, JDK 1.4.0_01 as well, by way faster and dual
CPU), and get consistently the same failures on asserts in lines 232 & 245!
When I comment out these two lines, the DateSelectorTest completes
successfully (as does the entire Ant testsuite). Not sure what's wrong. --DD
-----Original Message-----
From: Bruce Atherton [mailto:bruce@callenish.com]
Sent: Sunday, July 07, 2002 4:19 PM
To: Ant Developers List
Subject: Re: Problem with latest Selector patch?
At 10:29 AM 7/6/2002 -0500, Dominique Devienne wrote:
>Fresh sandbox of Ant branch 1.5 from this morning. Just did bootstrap and
>build test. Win2K, JDK 1.4.0_01. --DD
>
>Testcase:
>testSelectionBehaviour(org.apache.tools.ant.types.selectors.DateSelectorTes
t
>): FAILED
>expected:<TTFFTFFFTTTT> but was:<TFFFFFFFFFFT>
Ok, I've just done a number of tests with fresh checkouts and I can't
reproduce this error. With Win98, I have produced a different error in
org.apache.tools.ant.util.FileUtilsTest that looks similar, though. Here
are the details:
Win2K:
1) JDK 1.1.8 (all shells) - All selector tests pass (when basedir in
selector.xml is adjusted)
2) JDK 1.3.1 (all shells) - All pass
3) JDK 1.4.0_01
a) CMD - All pass
b) 4NT - All pass
c) Cygwin - org.apache.tools.ant.util.FileUtilsTest fails, line 405
Windows 98:
1) JDK 1.1.8 - All selector tests pass (when basedir in selector.xml
is adjusted)
2) JDK 1.4.0_01 - org.apache.tools.ant.util.FileUtilsTest fails, line
122
Linux (Debian Woody):
1) JDK 1.3.1 - All pass
2) JDK 1.4.0_01 - All pass
Just to be sure this isn't intermittent, I reran the test on Win2K with JDK
1.4.0_01 a dozen times. I ran within CMD, 4NT and Cygwin. I even tried
removing and refetching the expected files a few times, in case it was a
random error based on the timestamps of those, but it passed every time.
The line that is failing in the Windows 98 run is
"fu.setFileLastModified(removeThis, 123456);", which sets the last modified
time on a file to a specific millisecond value. That is interesting because
the line that is failing for Dominique is comparing for a specific
millisecond time as well - "s.setMillis(1006347300000L);". The Win2K Cygwin
failure doesn't appear related, as the error is "expected:<d:\bar> but
was:<D:\bar>".
So after all this testing, I have to say that I am at a loss as to what
Dominique is experiencing. I can comment out that particular test, but then
we won't be exercising the setMillis() method in DateSelector.
What is the right answer? Should I nix it anyway? Also, Dominique, if you
comment out line 232 of DateSelectorTest does everything else then work for
you?
--
To unsubscribe, e-mail: <mailto:ant-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:ant-dev-help@jakarta.apache.org>
--
To unsubscribe, e-mail: <mailto:ant-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:ant-dev-help@jakarta.apache.org>
|