Bug work

“People have forgotten this truth,” said Fox,
- but you do not forget:
you are forever responsible for all those who tamed ...

Antoine de Saint-Exupery "The Little Prince "

Only the one who does nothing is not mistaken. Over time, errors accumulate. What can I say, right now in my hands there are a dozen use cases that lead to various errors in the Quantum Noughts alone and are completely lacking in spirit to deal with (the code there is really hellish ). But sometimes I find time to fix something. And you know what? Correcting old mistakes is no less interesting than making new ones!

Old Ur in new clothes

Ur is one of my very first games (if not on Zillions of Games , then on Axiom for sure ). This is the game that started my passion for board games! Of course, this is not about the reconstructions of Bell or Murray . For all their virtues, these pundits, apparently seldom played board games (at least according to their own rules). The only, in my opinion, playable version of the rules for this game, was developed by Dmitry Skiryuk .
')

Fortunately, Nomad1 almost immediately joined the creative process, having developed an excellent implementation of the game under Android and iOS . It should be noted that another one of Habr's authors took part in the development. Without the excellent musical themes created by 1eqinfinity , the game would not have been so atmospheric in half. The resources of the Android version have significantly improved the appearance of the ZoG version, but one of the important points was not given to me in any way.


In the first version of Hurray, I hid the locked figures, imperceptibly moving them to invisible fields. When the top chip went away, I returned the pieces from the pile, one by one, to their original place. Unfortunately, visibility suffered. With such a “view from above,” it was necessary to have a good understanding of the rules in order to distinguish the situations of “knocking down” chips from their “covering”. More recently, I managed to solve this problem:



In Ur the figures do not move in groups. A single chip can be placed on top of the “stack” (in some fields of the board) or leave the top of the “stack”. In this regard, the rules of the game in “Ur” are ideal for the chosen approach. The only thing is to correctly determine the positions on the board (by moving a part of the positions up a couple of pixels up and to the left) and link them together:


The approach works, but in some cases the picture breaks:

In general, everything is clear here. We deal with rectangular tiles, partially painted with a “transparent” color, but when ZoG renders a “transparent” color, it takes a part of the board image (background image), and not a figure that overlaps the same place. I have already come across a similar when drawing shapes for " MarGo " and I know how to deal with it. All you need is to draw additional tiles with shaded lower right corners. The fill color must match the color of the figure below.

Unfortunately, the matter is not limited to the preparation of graphic resources. The fact is that the shell Zillions of Games "does not like" to perform "extra" work. When moving one shape, the others do not redraw. After performing the animation of the move, the figure is simply drawn at the new position, and the old position is filled (again) with a rectangular fragment of the background image of the board. There is a way to "deceive the system". It is necessary “only” to initiate the redrawing of all the figures, starting from the bottom of the pile:


In fact, all the “stack” pieces are re-created with the same types, as part of the move that removes the top chip. This is already a workable solution . Of course, if you closely follow the animation, you can see a brief destruction of the image when removing the chips or the animation of the movement of the figure with the filled in "corner", but at the end of the move the picture is restored to a completely acceptable state. Perhaps this is the best that can be achieved by means of Zillions of Games.


As a bonus, I received an unexpected gift. In general, I was planning to rewrite the Axiom-version after ZRF, but it was not needed! Because of the simplification of the logic of the game, AI began to cope with the game much better. Suffice to say that sometimes he wins from me, and in most cases, he does not have time to spend only one chip. It is still worse than the Android version, but it is already quite pleasant to play with it.

Hunting for the "Elephant"

It was not my mistake, but my hands were very much itchy to fix it. I do not know why. This is one pretty original chess-like game supposedly popular among the Ituri pygmies. According to the author , the description of the game was discovered by his friend in the diaries of the French missionary Maurice Morceau, dated 1821:

French French French French French French French French French French French French French French French French French French French French French uri His personal effects, including the diaries, were later found in a cannibal village by anthropological expedition.

Unfortunately, we are dealing only with the author's reconstruction. The game description was obviously incomplete, there was no access to the original source, and contact with a friend was subsequently lost. In particular, it is doubtful whether the use of a “ knight's move” movement in the game is too abstract for such a traditional game:

5x5 on author author author author author author 5 5 author 5 5 field 5 a Knight-move, which is sort of abstract. However, it’s impossible to hopefully win if colorbound.

By the way, this quote uses " Ralph Betza's '' funny notation '' ", which I wrote about earlier . Why am I so interested in this game? Let's try to figure it out.


We see one of the rare games that use figures of two different sizes. An elephant is a big figure occupying four fields of the board at once. It can move one “big” step along any orthogonal plane and “squeezes” everything that happens to be on the four target cells. He is opposed by the "pygmies", moving by the "Horse", as well as one "shaman", moving on one or two cells, on any of eight sides.

Of course, “small” figures cannot just move to cells occupied by an “elephant” (that would be too easy). An elephant can be “eaten” only under the condition that all the fields under it are attacked! The author himself draws parallels with the medieval European game called "The Fox and the Geese ". Not a bad idea, but, unfortunately, the implementation pumped up.


There are two mistakes here. One offensive, but the other - fundamental! First of all, quite unexpectedly, a shaman can attack a lonely-standing elephant (at the same time, quite expectedly, “breaking” the picture). At the same time, he does not eat the elephant itself, but simply occupies the cage below it:


The root of evil is here:

 (if empty? add ...) 

If the cell is empty - then we get up on it (and move on, but in a different reality). The trick is that the cell "under the elephant" is really empty. It is in another grid, small and “elephant” would not fit in it, with all the desire. This is pretty easy to fix:

 (define my-empty? (and not-friend? (or (empty? c) (vulnerable? c))) ) ... (if (my-empty?) (if (not-empty? c) (capture c)) add $2) ... 

Further more:


In such a simple way, using a shaman as bait, a lonely pygmy can defeat a whole elephant! Of course, this is a bit different from what the author intended. But why did this happen? In fact, everything is simple. An elephant can be taken if all four cells on which it stands are under combat. Now let's see the code:

 (set-attribute vulnerable? (and (attacked? ul) (attacked? ur) (attacked? dl) (attacked? dr))) 

Everything is quite transparent and it would even work if only the "pygmies" participated in the game. Performing a battle with the “knight's move”, they do not attack the squares of the 2x2 square on which they are located. Shaman - is another matter! It "covers" 3 of the 4 fields of the square on which it stands. It remains to “cover” the field on which the shaman himself is located and the trap is ready. And it does not matter that the shaman will be eaten - the attribute of the vulnerability will already be installed and the whole next move will work!

By understanding the error, you can easily fix it. The patch itself is quite verbose, but its essence is simple. We refuse to use the predicate attacked? in this game (it simply does not fit here) and is replaced by manual checks. A more flexible approach allows for special treatment for the "shaman". With the appearance of the game, unfortunately, nothing can be done. The picture constantly “breaks”. This game is not very suitable for Zillions of Games (but the idea itself is certainly interesting).

Do not look back!

In the process of working on new games, I know the Zillions of Games platform better. Complete new solutions come to my mind, which I couldn’t have thought of before. Often they are so good that they allow other older games to be fixed. So, in one of the articles by Dmitry Skiryuk, a very original game " Duel " is described, in which instead of chess pieces there are used dice.



When performing a move, the cube rolls across the board to the number of steps that was depicted on its upper face at the beginning of the turn. Movement is not performed in a straight line, the direction can be changed in an arbitrary manner. This is important because the face to which the cube will be turned upwards at the end of the turn will depend not only on the starting and ending point, but also on the chosen path of movement.

This kind of movement (with possible changes in direction at each step) in board games is quite common. This and Thunderclap in " Ko Shogi " and quiet moves of the figures in the " Nine-Tile Cyvasse " and much more. The main problem associated with it is the need to prohibit returning to previously completed fields. If the maximum distance does not exceed three steps (as in Claude Leroy’s Mana game), it is enough to prohibit the change of direction to the exact opposite (for this, the ZRF has a remarkable predicate not-last-from? ).


If the figure can move more than 3 steps, not-last-from? no longer saves. You must somehow mark all previously passed fields. In ZRF, for binding to fields of Boolean values, so-called “positional flags” are used. Unfortunately, they are not part of the game state and are automatically cleared at the beginning of each partial move.


Invisible figures can be placed on the visited fields (they are certainly part of the state). The main thing is not to forget to delete them in time! This technique is so versatile that it can be useful in a variety of completely different games. First of all, various transitional games such as Salta or Traverse come to mind, but the game is not limited to games of the Halma family. The same technique allowed to bring to the "working" state " Luzhanqi " and the long time ago written " Ossetian checkers ".

Instead of an epilog ...

Continuing the conversation about drafts, I want to note that the “work on the bugs” is not always reduced to purely technical issues. Very often, in the process of working on the game, I learn something new. Sometimes these “details” literally turn over all my ideas about the game, forcing me to look at it in a new way. As an example, I want to talk about one of the most interesting varieties of " Turkish drafts ".


This game is played in Bahrain . All the rules of "Turkish drafts" are executed. Only one new rule is introduced, which is absent in the original game. This rule is very simply formulated and, in some cases, significantly affects the character of the game, making it more combinational. For me it was a very unexpected and interesting find. Can you determine what it is, by recording the game?