Community
Participate
Working Groups
I20050426-1700 - have this code (caret at |): package codeAssist; import java.util.List ; public class Extends { void m() { List <? |> } } - invoke code assist > actual: the only proposed item is the CU's type (Extends) < expected: there are exactly two proposals: 'extends' and 'super' - alternatively, press 'e' and invoke code assist > actual: dozens of type proposals are proposed (and two template proposals, but that is not a jdt-core problem) < expected: only 'extends' is proposed.
Fixed and test added GenericsCompletionParserTest#test0186() -> test0187() CompletionTests_1_5#test0217() - test0220()
Build I20050526-2000-win32: - List <?|> and List <? |> ring a bell (no proposal at all, pipe figures insertion point while invoking code assist); this is at mismatch with the initial request, but consistent with code assist behavior in other places; I opened a separate bug (https://bugs.eclipse.org/bugs/show_bug.cgi?id=96918) to track this enhancement; - List <? e|> yields extends, plus a couple of 'else' templates (don't know if this is avoidable without qualifying templates, which may prove expensive/difficult to specify); note that no classes or interfaces show up, which is better than 3.1M7 behavior and the main subject of the bug; - List <? s|> yields super, plus a series of templates. Added tests passed successfully. I would then consider this as verified, modulo a new enhancement request for extends+super completion in the context of List<?|> or List<? >.
filed bug 96924 about the misleading templates - they should at least be ranked lower than the correct keywords.