On 8/11/06, Henri Yandell <[EMAIL PROTECTED]> wrote:
> On 11 Aug 2006 13:10:48 -0000, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> > Failed build logs:
> > http://people.apache.org/~psteitz/commons-nightlies/20060811/io.log
Brain dumping in case anyone else is looking at this too.
This error is in the isNewer test.
<snip/>
It creates a file, pauses for a second, creates another file, pauses
for another second and then creates another file. I suspect the bug is
that pausing for a second still ends up with a file with the same
timestamp for some reason or other. Easiest fix would be to pause for
a little longer than a second I suspect. It doesn't always happen.
<snap/>
The same spin method is used here to pause for a second. It looks like
it's overkill implementation wise; I thought you could trust
Thread.sleep(n) to pause for at least n milliseconds - unless an
InterruptedException happens (which in this case will fail the test).
Bumping up to 2000 surprisingly still saw errors happening. I also
don't get why the FileUtilsTestCase one is seen more often than the
FileFilter one.
<snip/>
Thats a troublemaker. Seems to be some timing subtlety on the
nightlies box that escapes me (being the original author of that
test).
-Rahul
Hen
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
