| 1 |
Things I want before KDE 4 beta 2 |
| 2 |
================================= |
| 3 |
[ ] Board is way to small at first startup. |
| 4 |
[ ] There is no indication the game starts in a demo mode |
| 5 |
[ ] Game over message is displayed in status bar. Not nice. |
| 6 |
[X] Remove all names from source files. For copyright holders, refer to svn |
| 7 |
log. We want collective code ownership. |
| 8 |
['] Documentation/Handbook. |
| 9 |
['] Game rules |
| 10 |
['] Game features / "Help" |
| 11 |
['] Game strategies |
| 12 |
['] Game history (Gomoku, Connect5, ...) |
| 13 |
[ ] Theme creation |
| 14 |
[ ] Apidox. |
| 15 |
[X] Ai |
| 16 |
[X] Game |
| 17 |
[ ] Gui |
| 18 |
[ ] Mainpage.dox |
| 19 |
[X] Fully i18n. |
| 20 |
['] Code beautification. |
| 21 |
[X] Ai |
| 22 |
[X] Game |
| 23 |
[ ] Gui |
| 24 |
[ ] Naming conventions |
| 25 |
[X] Indentation |
| 26 |
[X] close namespace with comments |
| 27 |
[X] close #endif with comments |
| 28 |
[ ] sort functions, methods and properties |
| 29 |
[X] name all #ifdefs __CLASNAME_H__ |
| 30 |
[ ] /* comment */ all private stuff |
| 31 |
[ ] Tooltips. |
| 32 |
[X] Theme support. |
| 33 |
[ ] Configuration dialogue. |
| 34 |
[X] Save AI level and theme from session to session. |
| 35 |
[X] Logic cleanups. |
| 36 |
[X] Allways use Player instead of int |
| 37 |
[X] Give the best AI level a different name |
| 38 |
[/] Get rid of exceptions, to fit into KDE (AI left) |
| 39 |
[ ] Implement draws (yes, it have never ever happened to me in real life, but it |
| 40 |
is not impossible in theory, ...right?) |
| 41 |
[ ] "New game" doesn't mean a loss if it's just started. |
| 42 |
[ ] Kidnap an artist for some decent artworks |
| 43 |
|
| 44 |
============================================================================= |
| 45 |
|
| 46 |
THEMES (4.0) |
| 47 |
============ |
| 48 |
[ ] scribble (pen on paper) [IMPROVE] |
| 49 |
[X] spacy |
| 50 |
[X] gomoku |
| 51 |
[ ] KDE vs Gnome |
| 52 |
[X] high contrast |
| 53 |
|
| 54 |
============================================================================= |
| 55 |
|
| 56 |
FEATURES (4.1) |
| 57 |
============== |
| 58 |
[ ] Status message and game events printed "onboard" |
| 59 |
[ ] configure dialogue |
| 60 |
[ ] keyboard navigation ("no mouse" gameplay) |
| 61 |
[ ] That nice new AI... |
| 62 |
[ ] KNewsStuff2 for themes |
| 63 |
[ ] Winning line fades in, move by move |
| 64 |
[ ] "Stains" support |
| 65 |
[ ] Rotate Stains by 90, 180 and 270 degrees |
| 66 |
[ ] "Stains" extended (paint all buttons, menus etc in pen-on-paper style) |
| 67 |
[-] Change replay & demo playback speed |
| 68 |
[ ] Change board size |
| 69 |
[ ] "Natural paper" whith coffee stains, etc. |
| 70 |
[-] Save a replay |
| 71 |
[ ] Open a replay |
| 72 |
[-] Save an active game and open it again |
| 73 |
[X] Start with latest game when opening |
| 74 |
[X] Save stats from time to time |
| 75 |
[ ] View history in a right or left pane, where previous moves are displayed. |
| 76 |
[ ] Enable LAN gaming (broadcast the local domain for other willing to |
| 77 |
participate.) |
| 78 |
Look into kgamelib or what it is called. Create XML-like open protocol for |
| 79 |
Bovo friends, letting Gnomes play with us (is it possible to add a feature |
| 80 |
to the protocol that allways makes them loose? =P ) |
| 81 |
|
| 82 |
ADAPTING AI |
| 83 |
=========== |
| 84 |
ADAPTING AI: AI skill starts at Normal, and then it gets reevaluated after |
| 85 |
every gameover. The reevaluation takes into account the last 10 games played (if |
| 86 |
10 games hasn't been played yet it's a new user and special rules will apply). |
| 87 |
However, this 10 latest games is "evaluated" (who won the most, the player or |
| 88 |
the AI?). If AI wins more than 4 out of 10 games, lower the difficulty by a few |
| 89 |
points on a 100-level scale. If player wins more than half 7 games out of 10 |
| 90 |
games, make the AI slightly more challenging. This will mean that the player (as |
| 91 |
he/she) gets better will start facing more challenging opponents, but still they |
| 92 |
won't be disshearteningly hard. The algorithm is meant to mean that a balance |
| 93 |
will occure were the player allways faces an opponent he will win ~60% of the |
| 94 |
games against, making him feel good about him/herself, lowering his/her blood |
| 95 |
pressure and making him/her like this game. :P |
| 96 |
[ ] Implement background cogitating in AI (thinking over possible plays in a |
| 97 |
separate thread while player is thinking). |
| 98 |
[ ] Implement forsight in AI (AI guesses what moves the player might attempt) |
| 99 |
[ ] Implement depth searching in AI (If player plays this, then I could play |
| 100 |
this and the player could do this and I can do this...). |
| 101 |
[ ] Implement width searching in AI (AI evaluates many possible moves; multiple |
| 102 |
depths) |
| 103 |
[ ] Make sure width+depth searches is reflected in the "root" of the AI, so the |
| 104 |
AI chooses the path that seems like the best one. AI will be able to |
| 105 |
deselect entirely paths it discovers leads to losses or strong |
| 106 |
playerpositions. |
| 107 |
[ ] Implement a board vurdering (count offensive and defensive points appart and |
| 108 |
see which points is valued the most, to guess which player is most likely to |
| 109 |
like a given board). |
| 110 |
[ ] When a move is made by the player, drop all know redundant paths from the |
| 111 |
strategy and focus on the ones left. |
| 112 |
[ ] Make the AI cogitating thread do multi-tasking. That is, all |
| 113 |
depth-width-searches gets added to a wait list, and the AI will in turn |
| 114 |
vurdere them. Removes from this list should be easy as well (certain losses, |
| 115 |
we want to avoid if the player is not of the weaker kind AND when the player |
| 116 |
plays a move, all children to other moves must be removed -- Actually |
| 117 |
it's better if we just clear eveything specified to be not in a certain |
| 118 |
path.) |
| 119 |
[ ] Make the aiboard reversable (return a copy and change the playedID) so it is |
| 120 |
easy to alter between player and AI "thinking". |
| 121 |
[ ] Make this super-AI adjustable, by making miss-calculations (only in |
| 122 |
AI-thinking -- it has to presume player is allways a mastermidn) and by |
| 123 |
selecting weaker moves (unless they are extremely obvious, like |
| 124 |
four-in-a-row). Also make the AI weaker by limiting how deep and wide it is |
| 125 |
allowed to calculate. |
| 126 |
[ ] Make the width of paths adjustable by how likely they are. If there are many |
| 127 |
similarly pointed moves to make, explore them all, but if there is only a |
| 128 |
few great only evaluate those. |
| 129 |
[ ] Set max depth and width. The closer to the root, the "wider" path is |
| 130 |
allowed, and the better paths can also be explored wider and deeper. |
| 131 |
[ ] Allow the AI a few seconds of cogitating before forcing it to make a move. |
| 132 |
[ ] Find a nice balancing algorithm that ensures the player will face |
| 133 |
slightly-easier opponents. |
| 134 |
[ ] If 10 games hasn't been played yet, start with a random difficulty. If that |
| 135 |
proves to be to hard, back down by half (and so on) or double the |
| 136 |
difficulty, until the players starts loosing/winning. Then back down |
| 137 |
halfways to the last value and continue to do so whenever the |
| 138 |
winning/loosing situation changes but keep making it harder/easier if the |
| 139 |
situation doesn't change. When 10 games has passed, start playing with |
| 140 |
normal adaption. |
| 141 |
[ ] Turn on highscore and connect that between adapted difficulty of AI and time |
| 142 |
of play. |
| 143 |
[ ] Turn on global highscore. Send in exceptionally great levels of difficulty a |
| 144 |
player has reached to an Internet server, so a top-100 masterminds of the |
| 145 |
KDE world list is created. :) |
| 146 |
[ ] If window gets inactive, pause the AI cogitating thread. |
| 147 |
[ ] Refactor out this cooooool adapting AI into a lib, which all KDE Board games |
| 148 |
could use! (GSoC idea?) |
| 149 |
|
| 150 |
BUGS |
| 151 |
==== |
| 152 |
[X] Several marks are lost when winning line is painted in replay mode. Odd.. |
| 153 |
[ ] This might be buggy, haven't checked: when starting an autosaved game; |
| 154 |
continue the computerbegins-playerbegins sequence (who made the first move in |
| 155 |
the history?) |
| 156 |
[ ] Area outside of scene isn't updated correctly when switching theme. |
| 157 |
[X] Wins vs. Losses is not updated on startup, but at first when one both has won |
| 158 |
and lost. |
| 159 |
[X] At startup, playbackSpeed is reset to 2000, due to it beeing to big. Why????? |
| 160 |
[X] Undo crashes. |
| 161 |
|
| 162 |
OTHER |
| 163 |
===== |
| 164 |
[ ] Move this list to a wiki! |
| 165 |
[X] Clean up the namespace mess. |
| 166 |
[ ] THE SAME THING WE DO EVERY NIGHT, KONQI: TRY TO TAKE OVER THE WORLD! |
| 167 |
|