Talk:Amb: Difference between revisions

What I think the first sentence is trying to say (and also the conventional meaning) is that Amb is a bottom-avoiding choice combinator. Suppose you have a function <math>f</math> that works fine for some inputs <math>x</math>, but crashes or fails to terminate for others. Similarly a function <math>g</math> works only for some limited set of inputs, not necessarily the same as <math>f</math>. Then <math>\mathrm{Amb}(f,g)</math> is a function that returns either <math>f(x)</math> or <math>g(x)</math> when given an input <math>x</math>, but is guaranteed (!) to choose the one that terminates correctly if one of them does and the other doesn't. Once you understand this definition, you'll realize that the only right way to implement Amb for arbitrary unknown functions <math>f</math> and <math>g</math> is by multi-threading or interleaving on some level, which is where the emphasis in this task belongs. (You can't get away with using only exception handlers unless you excludewant to neglect the possibility of non-termination.) Note that this semantics differs from non-deterministic choice. If <math>f</math> and <math>g</math> happen to have disjoint domains, it's completely deterministic.