> There is something more surprising at work in itoe, though. itoe was > intended to convert explicit lists into implicit lists. At one point, > it is considering the case that the implicit list was empty. If you > put the cursor in the shed "itoe (lcc eq nil oh') []" then horace > tells you the surprising fact that "oh' : So false". Really? Who > authorised horace to apply the So constructor to false? Why can't it > see that this case is a dead-end? (it seems to me similar to the vec > zero X cases, which epigram does understand to be dead ends).
Horace is dumb, it says oh' :So false because that's what it's been told from reading off the hypotheses of itoe, you have to point out the absurdity and the method is simple, do case analysis on oh'... As for the other problem, i agree with edwin... by writing a more informative contains type (in fact writing more informative types in general) you'll get much farther cheers Peter
