Tengo Copy. Legalities? Usefullness? Thoughts?

**oeri** · 08-21-2006, 02:05 PM

Originally Posted by pRoFiT

Okay for the programming side of it. So curently i have to search the file once per letter typed. Each read from the file returns a number and a word, Also an integer which i will use later for popularity of the word.

There's a big slowdown. Load the file into memory. Perhaps make an array that contains each word and its values.

So CdR, i was kind of on the same page. I was thinking of creating the dictionary database for each starting letter group i.e. qwert.dic yuiop.dic etc.

This won't work. What if there are words that are the exact same numbers but have different starting letters?

Originally Posted by CdRsKuLL

was having a little think whilst I drove.. if you also organised the text file lin order of length of word.. ie.. ape , apple , applecart then this would also help with the speed. It would be quite easy to write a little app that loads a list of words, alphas them, then sorts them into length and ouputs them to another txt file.. as you say you would have some words that are more popular than others.. maybe like you say indexing them with a rating 1 to 9.. so you would have 1and 5apple 5applecart etc.. so it would then show the 1's first etc.. this still requires some nice coding :-) cant wait to see what you come up with

This is also a good idea. Make separate arrays for each word length.

**pRoFiT** · 08-22-2006, 10:55 AM

Originally Posted by oeri

There's a big slowdown. Load the file into memory. Perhaps make an array that contains each word and its values.

Yah but 100,000 words into memory. I would do pointer based array but VB sucks for pointers. I'm tempted to install delphi and write it in that. Its pascal based and im better at pascal.

Originally Posted by oeri

This won't work. What if there are words that are the exact same numbers but have different starting letters?

It would sort by popularity of the word used. So if you use "the" alot then "the" would come up first. If it was the same popularity of another word then it would be alphabetical next. I was going to use an integer for that but desided a byte gives me 0-255. This should be enough.......i may have to average them somehow. like when you use a word it goes from 0 to 255 as long as nothing else in the lists is 255. Then just divide by 2 to place the next least popular ie 0, 127, 255.....hmm..i still need to think about how this will work.

Originally Posted by oeri

This is also a good idea. Make separate arrays for each word length.

Yes then i could have 6 arrays to pull data from. that would make it real quick. Not sure how much memory these arrays are going to take up.

**Playerbeat** · 08-22-2006, 11:28 AM

try looking to this
http://www.dcs.shef.ac.uk/research/ilash/Moby/

**shotgunefx** · 08-22-2006, 11:38 AM

There are various structures that would help. Have you though about a bsp tree for quick indexing?

You'd probably want order it by the "keyed" text sequence.

Take a peek at Berkley DB would be useful, scales great, is freely available for almost all platforms and has a BSP type record set built in. Lot's of languages have bindings for it.

**pRoFiT** · 08-23-2006, 12:29 AM

Thanks guys. I think my original plan was fast enough. Currently im using pictures on the screen for a nice look. I switched to buttons with pictures on them and my speed is as fast as i can click. Although this is after i recreated my database to 6 dictionary files. Each at 700k.

So i thought a little about this after i saw the speed increase. you guys thought reading a 3 meg file would slow the program down. First off most of the newer hard drives have 4 to 8 megs bufferd. Im only reading 2.7 megs. most likley the whole file is in the hard drive buffer. Second, and im not too sure on this because i dont remeber if MBs stands for mega bytes or bits a second, the hard drive has a read of 133mbs. So if it means megabits then divide by 8 = 16.6 Mbytes a second. so my 3 meg file divide by 16.6 seconds = .18 of a second to read the whole file. that should be quick enough. but now that i made it faster, 1/6, it reads like at .03 of a second.

Maybe i'm wrong, but if i am that would mean mega bytes a second and i dont even want to calculate that.

So sufice to say it runs really quick now. I'm working on the popularity of the word right now and i kinda messed it up. now when you try the word again you can only get that word.

and nothing else. So when i get that fixed i will put out another alpha teaser. With the speed increase.

**oeri** · 08-23-2006, 01:28 AM

Originally Posted by pRoFiT

So sufice to say it runs really quick now. I'm working on the popularity of the word right now and i kinda messed it up. now when you try the word again you can only get that word.

and nothing else. So when i get that fixed i will put out another alpha teaser. With the speed increase.

That's why you should back up your source every day

**pRoFiT** · 08-23-2006, 03:16 AM

So far its only like 230 lines of code. I could probably start from scratch and have it done in an hour or so. Maybe. Anyways i fixed it. I was putting my data back into the database at the wrong position also

So every word used was in the database twice. Plus the mess up i had was related butmore like a broken feature i was adding. Now it works. Although my bubble sort is slow

Its been so long since ive used sorting algorithyms. I think quicksort is fast. Anyways i bubble sort the favorate word to the top of the list and then the database is already alphabetical.

So i might back up what i have today

to my memory stick. Maybe i can get some free time like you oeri and work at work

Anyways maybe tomorrow i will have another alpha to show. Anyone know how to highlight one word in richtext. like bold text

**pRoFiT** · 08-23-2006, 04:00 AM

okay updated alpha 2

http://home.comcast.net/~cstrachn/quicktype.rar

shift key may cause crash. pictures for it are linked to e:\up.jpg and e:\down.jpg