Unreachable levels are not dropped

[wismill]

Since #20 the following:

xkb_keymap {
  xkb_keycodes { <rightalt> = 1; };
  xkb_compat {
      virtual_modifiers LevelThree = Mod5;
      interpret ISO_Level3_Shift + AnyOfOrNone(all) {
          virtualModifier = LevelThree;
          action = SetMods(mods = LevelThree);
      };
  };
  xkb_symbols {
      key <rightalt> {
          [ Alt_R, Meta_R ],
      };
      key <rightalt> {
          type = "ONE_LEVEL",
          [ ISO_Level3_Shift ],
      };
      modmap Mod1 { Meta_R };
  };
};

compiles to:

xkb_keymap {
    xkb_keycodes {
        minimum = 8;
        maximum = 255;

        indicator 1 = "DUMMY";

        <rightalt> = 1;
    };

    xkb_types {
        virtual_modifiers LevelThree = Mod5;

        type "ONE_LEVEL" {
            modifiers = None;
            level_name[Level1] = "Any";
            map[None] = Level1;
        };
    };

    xkb_symbols {
        key.repeat = true;

        key <rightalt> {
            repeat = false,
            type[Group1] = "ONE_LEVEL",
            symbols[Group1] = [ ISO_Level3_Shift, Meta_R ],
            actions[Group1] = [ SetMods(modifiers = Mod5), NoAction() ]
        };
    };
};

Why not dropping the second level with Meta_R completely when setting type = "ONE_LEVEL", if it is not the level is not reachable? I think it should not even trigger interprets.

[mahkoh]

Why not dropping the second level with Meta_R completely when setting type = "ONE_LEVEL", if it is not the level is not reachable?

I didn't do this because it was too disruptive for the existing tests which rely on being able to use an arbitrary number of levels.

[wismill]

Fair enough. If the extra levels are fully ignored (no interpret, no impact on state, etc.) expect at serialization, then I guess it does not affect interoperability with xkbcommon, does it? It would only generate warnings about dropped levels?