This matches all three test cases I examined which failed when testing movePointLeft or movePointRight
I did not look deeply at every single failure in the deep history though; the failures are a different test case in every run which is curious; I guess the order of tests is random. Does it fail-fast?
sophia-guo
commented
1 year ago
Deep history shows consistent failures in bigdecimal for jdk 20:
https://trss.adoptium.net/deepHistory?testId=640da3e717052c6715411dd9
Examining three of those failures, I noticed in the sources there is a if-case separating jdk version 13+:
E.g. https://github.com/adoptium/aqa-systemtest/blob/0eb9f57f198fe194c070295c261e8bafeaed08aa/openjdk.test.math/src/test.math/net/adoptopenjdk/test/bigdecimal/TestSuite017.java
Then looking at BigDecimal sources, I noticed a change for jdk20 that suggests we should expect the behavior in jdk20 to match jdk11 again:
Latest commit of file: https://github.com/openjdk/jdk20/blob/master/test/jdk/java/math/BigDecimal/MovePointTests.java Points to: https://bugs.openjdk.org/browse/JDK-8289260
This matches all three test cases I examined which failed when testing
movePointLeftormovePointRightI did not look deeply at every single failure in the deep history though; the failures are a different test case in every run which is curious; I guess the order of tests is random. Does it fail-fast?