First element of `token` array in a multi-morpheme lemma contains the reading for the whole lemma

[fasiha]

Example: MeCab splits "休ませたら" into three morphemes, which Ve combines into one lemma. Ve's JSON-ified output contains the following:

...
    "tokens": [
      {
        "raw": "休ま\t動詞,自立,*,*,五段・マ行,未然形,休む,ヤスマ,ヤスマ",
        "type": "parsed",
        "literal": "休ま",
        "pos": "動詞",
        "pos2": "自立",
        "pos3": "*",
        "pos4": "*",
        "inflection_type": "五段・マ行",
        "inflection_form": "未然形",
        "lemma": "休む",
        "reading": "ヤスマセタラ",
        "hatsuon": "ヤスマセタラ",
        "characters": "0..1"
      },
...

Note how the reading and hatsuon fields for the first element of tokens contains the reading for the entire lemma, i.e., "ヤスマセタラ" instead of just "ヤスマ" as MeCab has (e.g., in raw).

I'm working around this oddity to build furigana, just wondering if it's intentional?

[Kimtaro]

Hmm, no, this does indeed seem like a bug :)