There are two tests that historically have code that is shared between CLDR and ICU:
When ICU transitioned to JUnit, this stopped working. So for now, CLDR can't run those tests in its environment, and we are dependent on running the tests after integration with ICU.
The new LocaleMatcher has data-driven tests, which are then easy to port. We can try to get the Transliteration tests working in both environments, or try to have at least some subset of the tests be data-driven.
Design Docs >