search

kebernet / shortyz

Shortyz Crosswords
GNU General Public License v3.0
107 stars 56 forks source link

Non-alphabetic characters in Puzzles (NYT Sunday SEpt 27 2015) #8

Open migdalorguy opened 8 years ago

migdalorguy commented 8 years ago

Sent you a note about this elsewhere. NY Times Sunday Puzzle 9/27/15 utilized non-alphabetic characters in the puzzle (in this case, the ".) If this is going to be a trend, can anything be done to allow access to non-alphabetic characters in Shortyz?

kebernet commented 8 years ago

So this one is hard to do. Let me explain why:

Sometimes these are REBUS characters. Like ♣ -- the problem is knowing how these are encoded in the puzzle. Usually, the puzzles are based on ASCII characters (the oldest US English character set still in common usage), so ♣ is encoded as "C" for clubs. The problem is, I can't know that the 'C' in that box really means ♣.

For puzzles that just have ASCII characters like '.' or '2' this is easier, but the question becomes do I give people hints based on their keyboard settings? It seems like the answer should be "no". There should be some special action you have to hit to put a special non-ASCII character in, but I haven't solved this problem yet.

migdalorguy commented 8 years ago

Great explanation, Robert. Hope you can eventually find a way to do it. (Personally, I didn't care for the quotation marks in Sunday's NYT puzzle, but I have a feeling we can expect to see more like this, perhaps even deliberately done to make it harder for those doing it electronically? Yes, I used to be proud of doing the Sunday NYT crossword in pen, but time moves on.

vin commented 8 years ago

relevant: http://wordplay.blogs.nytimes.com/rebus/?_r=1

One idea may be to long-press a square, and then maybe select a menu item, as in one of the examples listed in that post. But this just speaks to the UI question, not to interpreting the encoding / author's intent.

eevee commented 8 years ago

I've noticed an uptick in this kind of puzzle, too. Maybe it's worth reaching out to NYT to see if they could at least encode the solution correctly?

For characters not on the keyboard, it'd be nice if you could recognize the C keypress but display it as a ♣. But the puzzle would have to mention that, and I don't think the Across Lite format even understands ♣ (since it's based on ASCII), let alone have a way to define synonyms. There are more modern formats like IPUZ that might be able to... if NYT feels like trying them out. :)

Relatedly, there was another recent puzzle where a lot of squares had two letters each. I ultimately gave up on it because I kept forgetting what the filled squares were supposed to contain.