JS Crossword - a crossword where the clue = eval(answer)
63 points by rebane2001
63 points by rebane2001
Ily lyra dot horse I look forward to whatever insanity you come up with every time :DDDDD <3
omg this is too hard :)
I was mildly disappointed to find the solution not unique. In tr:nth-child,td:nth-child notation, the following single cells are underconstrained: 2,6; 10,18; 13,14; 16,16; 18,8. There’s also at least one (probably more!) that has multiple approaches: I think 38-down was supposed to use this, but someone else who posted a finished screenshot used /h/s which works too).
Correctness validation sometimes failed or misbehaved on intersecting cells, when an edit fixed one and broke one or similar. You’d need to retype a character to get the green. The page reload replay recognises my 38-down as solved, but the epic ending replay doesn’t. But more significantly, 9-down was my last unsolved one, I’d solved 22-across in a way incompatible with it, but when I changed the intersecting cell to fix 9-down, it decided I was finished, “whoaw!!” and all, even though 22-across was now broken! Page reload replay recognised I was not done.
A couple of times I had to search the JS standard library for names matching a glob. Not sure how I’d have gone without a local copy of MDN I could do things like find -name "??x?x.html" to.
I still feel 29-down is cheating.
Great fun. You owe me a chunk of my morning. 🙂
Thank you for playing ^^!
not unique
This is on purpose, and the same goes for multiple approach solves. It's a hacker-y challenge, so I think it's fair to allow different clever solutions if one can come up with them.
Correctness validation sometimes failed or misbehaved on intersecting cells
Do you have a way to reliably repro this?
intersecting solve
Nice catch, probably some fun async js race condition, but hopefully pretty fixable.
I see some dark lines. What do they mean? Do they separate strings (something I've never seen in a crossword before)? Edit: definitely looks like it. Seems obvious now that I see the way it highlights relevant squares, but definitely tripped me up on the first clue.
Yep, it cuts off an answer. It's pretty standard in crosswords, or at least in the ones they make in my home country.
I don’t go out of my way to do crosswords, but I see them from time to time and didn’t recognise this pattern.
This was fantastic! It took a few hours and a couple breaks to mull things over, but I solved it. I love experiences like this (also a big fan of regex crosswords). Thanks for putting this together and sharing this!
I only made it about half way through, but I learned a few things along the way (JavaScript has a lot of cruft, to go along with the weird edge cases!). The long ones were too open-ended for me to even get started. My favorites: 1 down, 33 down.
Regardless, this is an impressive creation!
all solved! i did not expect to get as far as i did, but it took me a few hours. this snippet helped me a lot (without revealing too much): T2JqZWN0LmdldE93blByb3BlcnR5TmFtZXMod2luZG93KQo=
https://gir.st/tmp/jsxword.png 😎
thanks for the fun, @rebane2001!
I have one complaint: it looks like if an answer going across is correct, the squares are all green, even if one of the intersecting vertical answers is wrong (in that same row). It's highlighted correctly on the list of clues (or, rather, not highlighted), but it makes it so that the crossword shows a square as green, even when it's wrong. I feel like if a square is wrong in either direction it should show up as red.
Sites that hijack and interfere with the browser back button should be marked as spam
what os, browser, and version? my site does not interfere with the back button, but i also saw one other comment claiming this, so maybe it's a bug with a specific browser?
Edit: fixed, turns out some browsers consider an iframe.src assignment to be navigation
