Probably not the place to ask, but some of you might find this interesting enough to take pity on me.<p>I had an idea years back to build crosswords on the surface of a truncated icosahedron (<a href="https://i.imgur.com/qlLl434.jpg" rel="nofollow">https://i.imgur.com/qlLl434.jpg</a>) but I had no idea how to find sets of 5-letter words that would mesh like that. A very clever programmer friend of mine whipped something up in Perl over a weekend using hashes. Basically it pre-emptively runs through the dictionary and hashes all wildcard forms of a word back to the original word (l...s->leaks, etc), then at each stage of filling out the puzzle it just grabs the list of words that hash to the current pattern. Say you're looking for something that fits s..p. it would return [stops, swept, soupy] etc, then tries each recursively. If the list is empty, there is no solution at that level so it backs up and tries the next word at the previous level.<p>I learned enough Perl to be able to make the code able to accept different polyhedra. The program takes quite a while to run on larger models, like say the rhombicosidodecahedron, so I've been looking for ways to speed it up. My background is with assembly and embedded C so I'm sure my implementation of these hash lookups would be slower than Perl's. Are there any other, maybe compiled, languages that have fast hash lookups like this?