Tue Mar 16 1999 17:54: I now have a full working version of TATC. It takes about a minute to search through a megabyte of text with a given row length. This is a big improvement over the other method. I don't have a time for the other method, but this is a big improvement. The improvement comes from noting where all the occurances of the first letter of the key are and putting them into a list, then just using the characters on the list instead of going through the whole text again and again (cause it sounded good to me).
Unfortunately, it can't handle much more than a megabyte of text at a time (at least on fire) before it runs out of memory. No way around this, I think, as the Java VM is a resource hog (fire is running at 1.7 load average, sorry guys).
I have another idea to speed up the search, but I need a letter frequency chart. Hey, I have Perl and a bunch of text, I can just make a letter frequency chart.