HCatalog splits it's different components (core, hcatalog-pig-adaptor, webhcat, etc.) into separate build.xml files, all controlled by the top build.xml. A side effect of this is that when the user does "ant test" and the top build file is iterating through the tests, if one of the component's test targets fail then unit tests are not run for subsequent components. This means a failure in one component can mask issues in others.
test should run on all the components even when there is a failure in one of the components.