On 17/01/13 10:05, Claude Warren wrote:
On Thu, Jan 17, 2013 at 8:50 AM, Andy Seaborne <[email protected]> wrote:
On 15/01/13 16:00, Andy Seaborne wrote:
On 15/01/13 14:57, Apache Jenkins Server wrote:
See
<https://builds.apache.org/job/Jena__Development_Test/org.apache.jena$jena-core/398/>
This is failing in a randomized test - it runs a number of times looking
for a "false". If it does not find one in the run limit, it fails.
I will try to bump up the number of attempts a bit before the test
decides to give up (it does not fail for me on my local machine).
At this point, I'm inclined to try a few tweaks of the values and if the
numbers get too large (and the tests slow down), then not put the tests
on the critical release path.
Andy
Ugg - still unstable.
It looks like these tests (were testMatch) were not part of the main test
suite, which is why they weren't a problem before. So I have removed them;
specifically, change the class name so Test* does not find them and added
some notes about them in comments.
(Claude particularly) Does that make sense?
Andy
To be honest, I just copied the tests over. I am not certain what
they are testing or even if they make sense. Random failure sounds
like a test case issue more than a code issue so I think your approach
is correct. Please update the convert tests jira to add fixing these
tests.
Not sure they are fixable as such - I think they are some testing code
supposed to be run to support development, and not part of a general
build test suite. They seem to generate random cases and probe for
conditions; I guess attempting to find corner cases but without
guarantee there is a "right" answer.
Andy
