If the available dalvik source really is complete, the real difficulty is integrating it with the rest of the system, i.e. wrap the Maemo APIs into Android ones, which can get tricky, esp when it comes to ones that are undocumented and/or closed. I do think it would be slower overall because system processes/launching are done very differently in Android, plus of course the overhead of the wrappers and the occasional headache of replicating NDI. QtDalvik as such to me makes no sense - you don't get the advantages of either Qt or Dalvik, more like just the drawbacks of both.