[Insight-developers] itkTimeStampTest failures - Need help

[Tom Vercauteren]

Hi all,

Some time ago I have been working on an optimization of itkTimeStamp.
The objective was to avoid mutex locks by using atomic operations. To
make sure I wasn't doing anything wrong, I wrote a small unit test
called itkTimeStampTest.

The implemented solution seems to work well in most cases... In some
cases however, the test is failing. See for example:
http://www.cdash.org/CDash/testDetails.php?test=5264697&build=107576
http://www.cdash.org/CDash/testDetails.php?test=6302502&build=123972

Since I commited the unit test and optimization at the same time, I am
not even sure that the original implementation was always passing the
test. This implies that there might be a bug in my test.

The problem is that I cannot reproduce the test failure and haven't
spotted the bug by simple code reading. Since TimeStamps are crucial
to ITK, I believe that this bug is worth reviewing and that it should
be fixed before ITK 3.8.

Thank you for you help.

Best regards,
Tom

P.S: Is there an easy way from cdash to get a graph of failing tests
that would represent all the machines? I managed to get a per machine
graph but would like to get an aggregated information.