Community
Participate
Working Groups
HEAD The quick fix for a raw type warning should at least be as good as for an unresolved type. import java.util.*; public class Try { void foo(Stack<List<Integer>> stack) { List li= stack.pop(); Lost lo= stack.pop(); } } For the first type (List), the only quick fix is Infer Generic Type Arguments, and that infers List<?>. But for the second type (Lost), the first quick fix offers to change the type to List<Integer>, which is all we need. Bug 318433 should fix Infer Generic Type Arguments, but we should still enable the working quick fix that just fixes that problem. Beware of duplicate quick fixes when the file doesn't contain any compile errors (just warnings)!
Created attachment 189562 [details] Fix
> Bug 318433 should fix Infer Generic Type Arguments, but we should still enable > the working quick fix that just fixes that problem. Beware of duplicate quick > fixes when the file doesn't contain any compile errors (just warnings)! Didn't add this smartness for several reasons: - The "Add type arguments" quick fix only adds type arguments, but the new quick fix "Change type to ***" can also change the base type. - The "Add type arguments" quick fix also helps in other cases (e.g. unsafe method invocation on a raw variable). - For perceived UI stability, it's better to accept some redundancy than to hide one of the proposals in case they match. - It is not easy to find out what the "Add type arguments" quick fix will do, so the duplication is hard to detect. Fixed in HEAD.
