Here’s a walkthrough for puzzle 20, a small but tricky Masyu that hinges on an obscure technique that I’ve recently failed to explain well. Of course, I encourage everyone to try to figure it out on their own, maybe using the hints from the comments on the original post. Here’s the puzzle, spoilers after the break. 
Continue reading
Feels like anything can happen with coded puzzles… Here’s one that’s probably of medium difficulty as far as coded corals go, though the puzzle type seems to be inherently difficult.
Rules Map the letters to distinct positive numbers, and solve the resulting Coral puzzle: Shade some cells, such that all shaded cells are connected by edge, such that no 2-by-2-square is fully shaded, and such that all unshaded cells are connected to the puzzle border by edge. For clued rows and columns, the clues give the lengths of all connected blocks of shaded cells, in any order.
Or see the instruction booklet.
Here’s another Maximal Lengths puzzle.
Just to note that I’ve published the source files to the booklets for my 24h hour set, with the hope that that might be useful in the future to someone wanting to format their own puzzle set.
Also, I saw that Stack Exchange should be launching a puzzling Q&A site soon. That could well develop to be a useful resource.
Here’s a slightly different take on the Liar Diagonal Slitherlink.
Rules Draw a loop that travels horizontally, vertically or diagonally from point to point. Clue numbers that are not crossed by diagonals are equal to the number of adjacent horizontal and vertical segments used by the loop. Those that are crossed by diagonals are different from that number. Furthermore, in every row and column, there is exactly one diagonal segment, and that diagonal crosses a clue.
Here’s a Prime Place puzzle, another type from the upcoming GP round. Finally a type that forces me to deal with rendering irregular grids! I wonder if we’ll have a hex grid in one of the next rounds?
Rules Place a number between 1 and 4 into each cell, such that each number occurs exactly four times. The digit sum of each horizontal or vertical word must be prime. The digit sum of a diagonal word is prime precisely if that word is marked with a gray diagonal. Here, a word is a maximal straight line of connected cells; words may have length 1.
Or see the instruction booklet.
Here’s another GP preview puzzle, a nice and easy Maximal Lengths. (There’s a harder one in the queue.)
Rules Draw a loop that visits each cell, such that the longest loop segment in clued rows or columns is equal to the clue. (Length is equal to the number of crossed grid lines.)
Or see the instruction booklet.
The Czech round of the puzzle GP will take place next week, the instruction booklet has been posted. Here’s a practice puzzle for one of the types.
Rules Place numbers from 1 to 6 in some cells so that no number repeats within a row or column. For rows and columns that have clues given on the outside, these numbers correspond to all sums of blocks of adjacent digits within that line, in the correct order. Furthermore, draw a loop that visits all cells without a number, passing horizontally and vertically from cell centre to cell centre.
A while ago I alluded to some puzzle rendering project I was working on, today I want to give a small update. It’s a Haskell library called puzzle-draw, the source code is available on github. It’s in a state where it’s useful to me (I used it for the entire marathon set, for instance), not necessarily quite ready for public consumption.
If you just want to play around with it, I’ve hacked together a web demo, below I’ll give an overview and explain how to install the command line tool.
EDIT: The command line tool has since changed to not require cairo by default, see the comments.
Continue reading
Bram made one of these for the 24 hour puzzle marathon, see his set. This puzzle probably explains why the marathon one was only solved by Germans…
Rules Shade some cells to form an orthogonally connected wall that contains no 2×2 square. The white cells within an area must form a tetromino, and two tetrominos that share an edge must not be of the same type. (The O-tetromino is possible.)
