DynComp failures with tests/daikon-tests/RatPoly about daikon

[mernst]

I have modified tests/daikon-tests/Makefile to disable this test for now.

Reported by [email protected] on 2014-02-05 17:09:01

[mernst]

typo - should be "make dyncomp-diff"

Reported by [email protected] on 2014-02-05 17:13:28

[mernst]

Is the whole test disabled, or just the parts that are currently failing?  It might
be better to keep the test enabled and either change the goal files (keeping this bug
report as an indication that the goal file should be changed back) or disable just
the parts that are currently failing.  Otherwise, it will be possible to make other
breaking regressions that this test would have revealed.

Reported by michael.ernst on 2014-02-05 17:58:36

[mernst]

Per Michael's suggestion, I have updated the baselines with the 'bad' goals so that
while waiting for a fix we can continue to run all the tests.  Hence, I have enabled
this test case.

Reported by [email protected] on 2014-02-05 19:35:46

[mernst]

Reported by [email protected] on 2014-10-27 22:35:55

[markro49]

Interesting note: if you use --no-jdk then you get the expected result. In the past, I noticed something similar with java.util.Random, but have not been able to find a similar cause as yet.

[markro49]

Decided not to use the instrumented version of Random as the method generates values based on an initial seed value. This has the undesirable side effect of putting all the generated values in the same comparison set when they should be distinct.
fixed with e16e5ad