[norswap]

I'm fairly familiar with Prolog, from its operations & implementation, having also thaught it to students.

Something I never really grokked however is: when to reach for it?

It seems like a powerful tool for a certain class of problems, but somehow I never seem to stumble upon them.

An obvious candidate is problems that map clearly to solvers that can limp along with sub-state-of-the-art performance. Solving sudokus is the classical classroom example. But somehow again, never really ran into something that maps to that (I have worked mostly in compilers and distributed system, with a smattering of application programming (frontend/backend)).

Any ideas or anecdotes?