The tests this introduces are, I believe valid. The failing tests represent fundamental semantic problems with the approach we've taken, so I have left them unfixed so we can discuss what the heck to do about it. Nevertheless it's probably OK to merge them if you want to; any damage this reveals is already done in the branch.
The tests this introduces are, I believe valid. The failing tests represent fundamental semantic problems with the approach we've taken, so I have left them unfixed so we can discuss what the heck to do about it. Nevertheless it's probably OK to merge them if you want to; any damage this reveals is already done in the branch.