Re^5: Turning A Problem Upside Down

[BrowserUk]

The nearest wordlist I could find to that was this one. It takes a couple of seconds longer to index than my 90,000 word list, but the time taken to locate the words hardly changes at all as the main oparations remains ORing & AND NOTing 26 bitstrings (essentially O(1)), and then the final filtering:

C:\test>wc -l TWL06.txt
 178691 TWL06.txt

C:\test>790206.pl
apetpxl
ale
alp
alt
ape
apex
app
appel
apple
applet
apt
ate
axe
axel
axle
eat
eta
exalt
expat
lap
lappet
lat
late
latex
lax
lea
leap
leapt
lept
lepta
let
lex
pal
pale
palet
palp
pap
pat
pate
pax
pea
peal
peat
pelt
pep
pepla
pet
petal
plat
plate
plea
pleat
plex
tae
tael
tale
tap
tape
tax
tea
teal
tel
tela
tepa
tepal
Found 64 words in 0.23

You don't say what the "certain time frame" is, but presumably as this is intended for the human brain rather than computer, it is in the order of minutes rather than sub 1 second? Even if I feed in the entire alphabet--which means it must select all the words via the bitstrings and then filter out all those that contain duplicates. ie. worse case--it only takes just over 8 seconds:

C:\test>790206.pl >nul
abcdefghijklmnopqrstuvwxyz
Found 36409 words in 8.46

[Limbic~Region]

BrowserUk,
Youo don't say what the "certain time frame" is,...

The time limit is 2 minutes but the majority of the time in my code is with an artificial delay in Win32::GuiTest's SendKeys() to not overwhelm the game.

I have not used bitmasks in the past quite the way you have here. It obviously has a huge advantage over my approach when the given input length grows. Some day I will pick your brain on this. Thanks again.