Consider:
|
public void firstOrDefaultLocalVarParam() { |
|
makeHelper() |
|
.addSourceLines( |
|
"Test.java", |
|
"import org.jspecify.annotations.*;", |
|
"@NullMarked", |
|
"class Test {", |
|
" public interface List<E extends @Nullable Object> { boolean isEmpty(); E get(int index); }", |
|
" public static class Collections {", |
|
" public static <T extends @Nullable Object> List<T> singletonList(T element) {", |
|
" throw new UnsupportedOperationException();", |
|
" }", |
|
" }", |
|
" public static <U extends @Nullable Object> U firstOrDefault(List<U> list, U defaultValue) {", |
|
" return list.isEmpty() ? defaultValue : list.get(0);", |
|
" }", |
|
" static void use() {", |
|
" String x = null;", |
|
" // should infer T -> @Nullable String", |
|
" String result = firstOrDefault(Collections.singletonList(x), \"hello\");", |
|
" // BUG: Diagnostic contains: dereferenced expression result is @Nullable", |
|
" result.hashCode();", |
|
" }", |
|
"}") |
|
.doTest(); |
|
} |
Our generic method inference currently does not work correctly for this case, as it incorrectly assumes x is @NonNull based on its declaration. Instead, inference should get nullability of local variables based on what is computed by our flow-sensitive type refinement, which in this case would be @Nullable for x.
This is a subtle change, since dataflow analysis also needs to invoke inference (see #1262), and so we have a cyclic dependence; will have to manage this carefully.