Pak_Man

broken code

2005-06-20T12:40:00.000-04:00

It's the day of submission and it seems that the only version of my code that I have on hand is that which I tried to loop 4 times to assign a direction to all 4 ghosts. This broke the code completely. I'll find a working version or fix what I have and submit it to the following address:

http://www.glan.ca/~mitch/working/ai.java
http://www.glan.ca/~mitch/working/levelOne.java

I should have it up tonight, tomorrow at the latest.

EDIT: According to my computer, as of Monday, June 20, 2005, 8:10 PM, both files are up.

Tired and underpaid.

2005-06-19T02:03:00.000-04:00

So it's late, and 3/4 of the group are at Mitch's house. We are nearing completion of the game, but still have work to do, so we decided to meet at Mitch's house and work on it together. Sitting near me (Jared) are Mitch and Nate. Ben is to be getting here in about 8.5 hours. ((Oh, Ben just called and apparently he will actually be here in about half an hour)).

I have made a list of what we still must complete, and the list is somewhat long-ish. It's not overly long, however, and although a number of the tasks are somewhat time-consuming, none of them are really all that difficult. So after that list, I feel somewhat reassured. Crunch-time is slightly less crunchy. Like an Oreo in milk.

Also, the computer I am using is HORRIBLE. Yes, it is even worse than my own. The video card can reach nothing better than 640x480 resolution, and the colour quality can not best 16 bit. The mouse is completely screwed, and if I move it, it goes wonky all over the screen and starts selecting things. So I use the keyboard. In JCreator, I cannot actually type, so I must type things into Notepad and then copy them. The ONLY reason I am willing to use this machine is because I need to code and it is the only thing available. :/

Anyway, we stil have work to do. But we are nearing our goal, and taking a break. We are all very tired at this point in time, and decided that we can get no productive work done and will therefore rest for some time, then get back to work when our minds are refreshed. And thus shall end.

*falls asleep*
~Genome

Aaaaaaand I'm spent.

2005-06-15T18:57:00.000-04:00

Yup, I'm done. That's it. Nothing left to do. Fully functional code on my end, in every respect.

Well, that might be a slight exaggeration, but pretty close. I've got a little bit more commenting to do, but random for ghost 0 works now. I've broken up most of the code into methods, too, so it's ultra clean, made more so by my re-organization of the variables. Other than that, I just need the class names.... and classes themselves that the x and y co-ords for the ghosts and pak_man are going to be pulled from so I can add that ability to my code (hardcoded for testing purposes at the moment).

So lets review, my code can:

determine the shortest straight line distance to pak_man if it were to move in any given direction
Do the opposite when pak_man is chasing the ghosts
If it's for ghost 0, it goes in a random direction

And it does all that regardless of the following pitfalls:

More than one direction holds the same distance
All distances are different
the ghost is one tile away from pak_man

The only possible other feature I could add would be the ability for the smartest ghost to look more than just one tile ahead. That, however, is not the most pressing issue when we don't even know how to get an image to move across the screen when the user presses an arrow key. In light of this, I feel the best course is to put it to a group vote tomorrow in class; whether I add the aforementioned feature, or help with the swing coding. If it's decided that I'm to work on swing, I'll spend 20-30 minutes commenting, and then my code is as good as ready to be submitted for marking.

I are teh l33t.

2005-06-08T22:45:00.000-04:00

Data transmission works. Checked over two computers, tomorrow I shall try three in the beginning of the class.

w00t.

Success! And problematic findings. :(

2005-06-07T17:35:00.000-04:00

Socketman here. With my Thread of Truth I shall conquer evil. (I really hope that somebody realizes that blatant Wonder Woman reference.)

Anyway, here's the update in the world of networking: As I said earlier, I believe, I have read and understood everything concerning multi-threaded "servering" and all was well. Now, I have completed the Server program (rather, the classes that will receive and interpret information from networked computers) and it works like unambiguity incarnate. Unfortunately, I had to include Swing to make it work. So in lieu of this, after testing is complete, I will simply include the component that requires Swing in Nate's section, as it is essentially just the part that calls my sockets and threads into existence. So it is easily transferable.

Tonight I plan on working steadily on the Client program (or, the classes that are intended to send information over the network to Server applications on receiving terminals) and hopefully I shall complete it tonight. After that, all that I must do is clean it up, determine which pieces of data must be sent through the network, and then finally determine how to specifically connect the computers.

The problem inherent in connecting the computers is a simple one, but with a seemingly difficult solution. Originally, my main problem was how to connect -all- the computers in every way. For instance, each computer must be connected to every other computer in order to transmit its data. I almost decided to have the information travel in a ring-like sequence, with each system connecting to only one, which in turn connected to another, etc. This would allow all information to be shoved around. However, it is terribly inefficient as the first computer's variables will take longer than necessary to get to the last computer. So I have now decided to have 4 Clients per terminal connecting to the Server running on each other computer. This is a much easier and quicker approach to my original problem.

However! My new issue arises out of this decision. For the way that sockets interact is using the Computer Name defined in network properties. Therefore, either the Client that is currently being run must KNOW the Computer Name of the Server it's connecting to, or the end-user must so that they can type it in. If need be, I can simply do that, and have a function to read the name so that each user can read it to the others to type in, but that is both horribly inefficient and somewhat dangerous if this game is ever to be introduced to those who are not so adept in the workings of computers. For this reason I must exert much contemplation on how to do this... Although I don't know of any other way.

We shall see what will come of this. I will speak with the others about it. Farewell for now.

*a string of a sticky web-like substance shoots down and pulls him away*
~Genome

cough cough

2005-06-06T11:50:00.000-04:00

Jared posting here.

It seems to be working. Don't know what's wrong.
I may or may not look into this later.
Probably not.

It may have been some sort of jump/glytch/whatever in the server for a moment. Hopefully it doesn't happen again, eh?

The Many Faces of Pak_man

2005-06-05T22:27:00.000-04:00

On friday I completed the 16 models of pak_man sized, colourized and ready for use. I have the normal pak_man's 8 different models; up, down, left and right for open and closed mouth positions. Also we decided to make, instead of the ghosts changing, pak_man changes when he eats the pill. So all 8 models for the "on the pill pak_man" are done as well. So the next task which I attempted on friday is to find a high-res format and low size (which shouldn't be a problem on a 22x22 object) to export it as. I have yet to find a file type that keeps the resolution but my quest continues.
~fruitsnacks

A map to success

2005-06-04T10:35:00.000-04:00

Given my accomplishment last night of perfecting the pathfinding aspect of the AI, I thought I'd give myself a checklist of things yet to be done, which is actualy quite a lot.

Pretty up the code: I've done a few things in quick 'n dirty ways, just to get them done for proof of concept that needs to be done better, as well as consolidating the variables to one place in the code and calling the making the variables that will call their data from different classes do just that
Further comment the code: rapid advancement like last night leads to poor commenting
Reversing the AI: Obviously if pak_man is chasing the ghosts to eat them, they can't be chasing him :P
Programming the different levels of AI: I still need to make the random AI for the stupidest ghost as well as the advanced AI that can look one or more tiles ahead for the smartest ghost.

As you can see, there is still MUCH to be done. I think the most important thing to be done is working out the reverse AI, which shouldn't be too hard. After that, if all else fails, I will have done the minimum of what I had set out to do, and my share of the work will at least be working enough to be playable. The advanced AI will have to be time allowing.

Very near complete victory!

2005-06-03T14:52:00.000-04:00

See title.

Putting more 'i' in 'ai'

2005-06-03T13:35:00.000-04:00

Majour breakthrough today in class. if two tiles give the same distance to pak_man, the ai works. On the otherhand, if all 4 distances are not equal, the ai breaks down.... But as I write this while waiting for Eclipse to download, I just realized that that is entirely fine. I'll just remove the part of that determins the direction if the distances are different and instead, after the part of the code that works, if it hasn't come up with a solution, I'll sort the distances from shortest to longest and choose the shortest. It's all coming together now....

Swing

2005-06-02T22:10:00.000-04:00

Swing is evil.

I am still attempting to learn it. EVIL. I have learned that I need to make a JPanel before using image icons, and I'm currently examining some swing code I found on-line to try to go back through it and see what makes it tick. I've been looking on-line and in the Java books in the classroom.

But I despise Swing already.

Until next time,
Wombat? What wombat?
~Kataron

Following a Different Thread

2005-06-02T19:41:00.000-04:00

So I have spent much time poring over the code that I acquired from that good friend of mine, and I understood it. The problem was that it was source with a purpose. It interacted with numerous other classes, ones which I did not have. That wasn't such a big deal, but it made it more difficult to understand the basics of sockets.

So I started from scratch (while still remembering what I understood) and found a tutorial on Sun's website. I have just been going through it, and am understanding wonderfully. The best part, I find, is that I have been analysing and compiling example code and I have now learned how to interact over a network with sockets perfectly fine. The thing is, I have only learned sockets. Not threads. So I can only interact between two computers at a time.

Not to worry, however. I have already begun learning multi-threaded sockets, and I do not believe it will be very much more difficult than single-threaded. Expect another new post from me very soon. In the mean-time...

*distracts you as he escapes behind a RETURN statement*
~Genome

The network is getting there...

2005-06-01T00:12:00.000-04:00

So I've worked on the network code a bit. Not much, yet... I am still working on understanding completely. I've also been trying to think of a port to run this on, at least for testing. Right now my mind is set upon 510, but that is not yet final. I recall adding the IP of my computer in the classroom (10.5.16.175) into the ConnectionManager, and I will add Ben's (10.5.16.144) into the playerConnection, to test connecting into the game from Ben's. I have yet to do this testing, for the classes are not ready... But I am setting it up. Anyway, I believe I can readily send information through the network... All I need to do is complete the actual connecting part, and then figure out how to run Threads so that lots of data can be sent and received simultaneously. Oh, the excitement.

The colours of success

2005-05-30T12:56:00.000-04:00

Well now that the level design is complete I will be beginning the colourization of the game. The level is the first task, colouring the walls etc. Then pak_man will be next and the ghosts. After this has been completed, I will be working with Nate to integrate them into the game. The next task will be to make the interface menu. The decision whether or not to make the menu graphical or text has not been reached, but the latest polls indicate a majority towards graphical, which in turn will be a burden. Morale is good, and so I continue...
~fruitsnacks

sigh

2005-05-26T14:52:00.000-04:00

Alright, well... The code I have was fun to work on and all... but there are too many exceptions to code in, and it's already damn ugly, so it's time to start fresh.

A* (pronounced "a star") is the form of AI that was used in the original pac man, and is still used as one of the leading forms of pathfind AI... But it's complicated. So, in loo of that, using an idea I gleemed from some intro pages on A*, I've decided to do the following:

A 2-D array containing every tile used at what co-ord. The ghost will look to this array for what tiles are around it. If it's any tile other than the path, it discounts it as a possible direction to go in. If it is a path tile, then it will take the co-ord that it would be at if it was on that tile and draws a straight line (so to speak) to pak_man. If that distance is shorter than the distance from any of the other (up to) 3 directions, then it will choose that tile to move to next. If 2 directions give the same results, it will randomly choose one.

This is a very simple way of doing the AI, with many glarring faults. The worst being that it is VERY easy for the AI to get stuck behind a wall, going back and forth since it can not look more than one tile ahead. However, it will be fairly simple to code and I know it will work without much problems. AND IT WON'T BE 11 IF STATEMENTS LONG! :P

After I get the code for this working, which probably won't take more than 3 classes, I'll look back into learning A*. If I can come up with working A* AI, then I will have one ghost using random, the middle two using this, and the hardest using A*. If after further consideration I find A* totaly out of reach for the time I have left, I will looking into advancing this AI so it can look more than one tile ahead... which may not be too hard. Or I could just have one do random, the next look one tile ahead, the next two, and the best 3 or more... Once I make it so it can look 2 tiles ahead, any more would be trivial to impliment...

Hmmm....

Double-time the threadhood.

2005-05-26T13:09:00.000-04:00

So I am currently attempting to work on two different classes.

I began one, being the playerConnection class, which is to act as the thread with which each player is to connect into the game. Then I realized how much longer that part was going to be, and switched over to ConnectionManager, which is to handle the sockets being used and manage the input and output to the player. Essentially the playerConnection (or thread) is to work through the ConnectionManager in order to interact with the other terminals.

This task is proving to be rather challenging, as sockets and threads are a whole new ordeal for me. I admit... I plan on poring over code and analysing it in order to find solutions and breakthroughs during the prom itself.

*flies away into the purple skies and through the candy clouds*

~Genome

Level Complete

2005-05-24T23:30:00.000-04:00

I have finally completed the detailed draft drawing of the level, all dimensions are exact and size of blocks are all worked out. Tomorrow the draft will be put into a vector format and digitized. Further research will ensue for level 2 along with the client's perspective and take on the design. I hope not too much repremanding will need to be done.
~fruitsnacks

Level Design

2005-05-24T15:46:00.000-04:00

I am having some troublesome obstacles with the level as I still try to find the correct dimensions through trial and error. I am near completion though of level 1. The pak_men and level 2 will hopefully ensue.
~fruitsnacks

Networking of the Masses

2005-05-19T17:42:00.000-04:00

BwaHA! I hath returned.

So, I spoke with said l33t programmer friend a number of days ago, and he sent me a couple classes that work with sockets and networking and the like. So I have been spending the past few days just reading them and analysing them, and I have now completed the act.

I plan on reading them again, quicker this time, just to get a better understanding. But I believe that it is about the time for me to start working out the code required to interact the five computers in the network. Interacting with two will be relatively simple, and in supposed foreknowledge it does not appear as though five computers will be much more difficult.

So, I have yet to actually hardcode anything, and I likely won't have a chance tonight because of another final project, but the source for networking should be started very soon, I believe.

*falls into a mysterious hole*
~Genome

One More Time

2005-05-18T21:51:00.000-04:00

It has come to my attention that I am the only group member that has not yet posted on here. So here it is, folks. Savour the moment.

I'm doing the general code. Right now, my main goal is to get an image to appear without using applets, so I'm learning the swing command. This should let me learn the things I need to. Yes. So I am learning it. Unfortunately, the Java Short Course is not short by any meaning of the word. Yeah...So I'm going to be reading through that and learning the swing command, and I'm going to start with a program to put an image on the screen, than I'll alter the code to make the image move, then I'll add boundaries for the image, then I'll have it so that the image changes as it moves. Thus is my plan.

And thus have I posted.

Until next time,
KHAAAAAAAAAAAAAAAAAAAAAAAAAAAN!
~Kataron

Graphic Desgin

2005-05-18T13:00:00.000-04:00

Today I have started to make the first level of the game in vector format along with some concept art of the pak_man characters. Level creation will be at a speedier process in the future once the grid is complete and block sections have been allocated size on the grid along with pak_man himself.

Variables

2005-05-18T12:41:00.000-04:00

I'm going to keep this post updated with all our cross class variables.

gstx[] //array of all 0-3 ghost's x co-ord
gsty[] //array of all 0-3 ghost's y co-ord

pakx //pak_man's x co-ord
paky //pak_man's y co-ord

gstdir[] //Array of all 0-3 ghost's direction. This is the resulting data from the AI class, given to the movement class

/*
gstnum //Which ghost the ai class will deal with. To be retrieved from the movement class.
*/
// Mitch, does this actually have to be cross-class?

pakStatus // either boolean or String. boolean for whether or not he can eat ghosts. String, depending on how we are to include different versions of Pak_man

more to come...

Upon further consideration....

2005-05-17T12:23:00.000-04:00

It could have worked, but not the best solution...

Here's what I'm thinking now:

intarray[][][] //3d array containing the x and y loc of all intersections, with the third space
used to contain a number from 0 to 7 denoting each type of intersection,
i.e. can go straight only, can go left only, etc. with dead end being 0
/* Intersection types:
0 = dead end
1 = forward
2 = left
3 = right
4 = forward, left
5 = forward, right
6 = left, right
7 = forward, left, right
*/

If gst loc = loc of intersection then
if intersection type = 0,1,2,3 then
blah, blah ... too simple to pseudo code

else if intersection type = 4 then
method to determin if forward
method to determin if left
etc etc.

Of course, I just now realise the it should be up, down, left, and right on the grid, not forward, left right, backward based on the perception of something on the grid.... which adds more intersection options.

The methods to determin up, left, right, down will have to do with subtracting the x of pak and ghost and the y. the methods will be referenced in the overall if statement that chooses the path for the ghost based on what intersection type it's at, thus reducing the amount of actualy code in more than one ways. I have the basic idea on how those methods will work, and will write the pseudo code for them later.

And thus, it begins.

2005-05-17T00:37:00.000-04:00

After over six hours of persisting and prancing about in such a confined realm of electrons and images, I have completed the most difficult task of all those ahead of us. Lo and behold! for before you there can be found the fruits of my labour.

I suppose there is a very small something to give credit to tonight, as otherwise I will have nothing more to write into this post. :)

We have, this evening, taken the general workload of conceiving and creating this project and divided it into four major section: GUI/graphics, network interface source, AI source, and interaction source.

1) Ben Mossman is going to be taking on the role of graphics designer, as he has the most experience and skill in this field. He is going to draw each sprite for Pak-man's movement, as well as the maps for various levels, the Ghosts, etc. Ben is also likely going to be our humour-provider in times of dire need.

2) The network interface is going to be coded by none other than myself, Jared Tyler. I plan on using the resources of the Internet, and potentially a hardcopy manual or two, in order to better understand the inner complexities of transfering data between two or more computers in a network. Along with this, I will also be learning from a l33t programmer friend of mine, as he has had much experience in the world of Threads. Then... I will put my newfound knowledge to work and unprecedented beauty shall unfold!

3) Mitch Pippard, known in some circles as "That ADD Kid", will be gracing us with his ingenuity with Artificial Intelligence. Rather than the presently well-known use of "sectors" to create the illusion of a Ghost's AI, That ADD Kid is going to make use of his mathematical skillz. He is going to implement Pythagorean's Theorem, comparing the coordinates of both Ghost and Pak-Man. Using this, the Ghosts will have an all-new form of AI, and he may also give each Ghost a different level of intelligence, thus deciding how difficult it will be to escape said Ghost's lengthy arm of swirling spirit. Mitch will also eat. We have a genius on our hands, and he must be fed.

4) And lastly, as always, is Nate Morse. Our ever-present and eternally wired colleague, who likes to call himself "Kataron", will be slipping himself into the pools of intermediacy. Nate is going to program the unbearably essential functions that are to interact networking with graphics, graphics with AI, and so forth. He must delve into the depths of discovery in order to learn what is necessary to print images onto the screen, allow the images to interact with one another, and force the various other functions, such as those pertaining to the networked systems, to immediately control what sorts of graphical beings that must be controlled by opponent players. Kataron's duty in this undertaking, although seems to be basic at first glance, is actually one of the most important, and possibly most time-consuming. But we all know Nate, and we all know that he can discpline himself to get it done... Maybe.

And that, dear reader, is what is currently going on with Team Awesome in the ICS4M class of John F. Ross. Check in often to stay up to date on our findings and accomplishments, as we are being held under lock and key, threatened with ball and chain or sometimes lance, until we have fulfilled the wish for a massive amount of documentation and understanding. Please send for help.

And for future notice, not all posts will be this long. :) This one was just the beginning, and therefore had a lot to cover.

I fare thee well!
~vanishes~