SpellBinding News

SpellBinding is holding up nicely.

Small Updates

I made a couple of minor updates. To improve usability on my phone, I eliminated the first-level header. In Lagrange, the site name (SpellBinding, surprisingly) appears at the top anyway, and now, the game is immediately at the top, so the keyboard does not obscure it. Works well on my android phone anyway.

I got tired of typing /msg blah.. on the phone keybord to report missing words to myself while testing and having a drink at the Regal Beagle. So I addes a simpler way to do that - just precede the entry with a period, and it will log/send me a message.

A truly annoying bug eliminated...

Emacs, annoyingly, eliminates the trailing space if a newline follows, and my game format requires each solution word to have a terminating space. This caused problems - eliminating the final word from the solution set, or occasionally (like yesterday) crashing the game. I finally got around to fixing that.

Full Auto next month

Starting next month I am going to try to go full auto with SpellBinding puzzles. That is, each puzzle will be randomly generated from the dictionary, as opposed to the current, error-prone method of generating puzzles and scheduling them manually, after extensive checking.

I am excited, as I will now be just a regular guy, solving the puzzles alongside of you, and bitching about the missing words!

Dictionary work

The dictionary is always a work in progress. It currently stands at 42K words and growing.

It is a fine balance as too many obscure words make for painful puzzles, but it is very annoying when the puzzle rejects words that you know!.

I am now in posession of the entire word index of the Merriam-Webster's dictionary, my dictionary of choice for resolving issues with words. It's about 90K (SpellBinding-appropriate) words, but is not usable by itself, as it contains many Scottish, British, and generally dialectal words that are against my rules.

However it is a great resource, as it allows me to reason about words in my dictionary!

For instance, I collected all verb roots (a verb for my purposes is a word in the dictionary that has -ed and -ing forms, simplified). I then conjectured that many of these have an -er noun form -- 'sniff' has 'sniffer', etc. I quickly extracted the potential -er words, but roughly half of them are not words.

Previously, I had to manually check each of those -er words at the web site, and copy into the dictionary. It was slow, and it took me months to get through half of them in spare time. Now, I can just find the intersection of my -er list and the entire MW dictionary -- conveniently using the Common Lisp (intersection ...) function!

And so I imported all 600 or so -er words, and it took me about a minute...

=>. index

=>.. home