\documentclass[10pt]{article} \usepackage{a4} \usepackage{epsfig} \usepackage{listings} \usepackage{tabularx} \lstset{language=Delphi}% \lstset{basicstyle=\sffamily\small}% \lstset{commentstyle=\itshape}% \lstset{keywordstyle=\bfseries}% %\lstset{blankstring=true}% \newcommand{\file}[1]{\textsf{#1}} \newcommand{\var}[1]{\texttt{#1}} \usepackage[pdftex]{hyperref} \newif\ifpdf \ifx\pdfoutput\undefined \pdffalse \else \pdfoutput=1 \pdftrue \fi \begin{document} \title{Programming GTK in Free Pascal: Using GDK} \author{Florian Kl\"ampfl\\and\\Micha\"el Van Canneyt} \date{July 2001} \maketitle \section{Introduction} In this article, some of the graphics primitives from the gdk toolkit will be demonstrated in a small game - breakout. The GTK toolkit widgets are built upon the GDK: Graphics Drawing Kit. The GDK does not know anything about buttons, menus checkboxes and so on. Instead, it knows how to create windows, draw on them, handle mouse clicks and keypresses. This functionality is used by the GTK widget set to create usable widgets. Sometimes, the widgets offered by GTK are not enough, and one has to fall back on the graphics functionality of the GDK to be able to do what is needed for a program. Fortunately, it is not necessary to create a GTK window and handle all GDK events to be able to use the GDK functions. The GTK widget set has a special widget, which can be used to draw upon. This widget is the \var{TGtkDrawingArea} widget. The use of the \var{TGtkDrawingArea} is what will be explained below. The GDK graphics functions will be explained using a simple arcade game, to demonstrate that the speed of the GDK is sufficient for the creation of simple games. The breakout game is chosen because it is conceptually simple, requires moving graphics and can be extended in many ways. \section{The drawing area widget} The drawing area widget (\var{TGTKDrawingArea}) is a simple widget which just provides a drawing window. It responds to all widget events, and adds additionally the 'configure\_event', which is called when the widget is realized (i.e. when the window handle is created.) The widget has only 1 method: \var{gtk\_drawing\_area\_size}, which sets the size of the drawing area. It is defined as follows: \begin{verbatim} procedure gtk_drawing_area_size(Area:PGtkDrawingArea; width:gint;height:gint) \end{verbatim} The arguments to this function are self-explaining. To use the drawing area widget, one should respond to the 'expose\_event'. This event is triggered whenever a part of the window that was invisible, becomes visible. The event handler gets an \var{PGDKEventExpose} parameter, which describes which area was exposed. This can be used for optimization purposes. To draw in the drawing area widget, the \var{Window} field of the \var{TGTKWidget} parent can be used. This is of type \var{TGDKWindow}. All drawing functions require a parameter of type \var{TGdkDrawable} which can be one of the \var{TGdkWindow} or \var{TGdkPixMap} types. \section{Graphics contexts} Most drawing functions do not only require a drawable to draw on, they also require a {\em Graphics Context}. A graphics context is a series of parameters that determine how lines are drawn, what colors and font are used etc. The Graphics Context is an opaque record, and its members cannot be accessed. The relevant parameters are set in a \var{TGdkGCValues} record, which is defined as follows: \begin{verbatim} foreground : TGdkColor; background : TGdkColor; font : PGdkFont; thefunction : TGdkfunction; fill : TGdkFill; tile : PGdkPixmap; stipple : PGdkPixmap; clip_mask : PGdkPixmap; subwindow_mode : TGdkSubwindowMode; ts_x_origin : gint; ts_y_origin : gint; clip_x_origin : gint; clip_y_origin : gint; graphics_exposures : gint; line_width : gint; line_style : TGdkLineStyle; cap_style : TGdkCapStyle; join_style : TGdkJoinStyle; \end{verbatim} The \var{ForeGround} and \var{Background} parameters determine the foreground and background colors. \var{Font} is the default font. The \var{Fill} field describes how areas are filled. It can be one of the following: \begin{description} \item[GDK\_SOLID] fill with the foreground color. \item[GDK\_TILED] Use the pixmap specified in \var{Tile} to fill the area. \item[GDK\_STIPPLED] Use the pixmap specified in \var{Stipple} to draw pixels that are in the bitmap in the foreground color. Other bits are not drawn. \item[GDK\_OPAQUE\_STIPPLED] Same as \var{GDK\_STIPPLED} except that bits not in the pixmap will be drawn in the background color. \end{description} The \var{clip\_bitmap} is used to define a clip area. The \var{ts\_x\_origin} and \var{ts\_y\_origin} define the stipple or tile origin. The \var{clip\_x\_origin} and \var{clip\_y\_origin} fields define the origin of the clipping region. \var{LineWidth} is the linewidth used when drawing lines. \var{Line\_Style} determines how dashed lines are drawn. It can have one of the following values: \begin{description} \item[GDK\_LINE\_SOLID] Lines are drawn solid. \item[GDK\_LINE\_ON\_OFF\_DASH] Even segments are drawn, odd segments are not. \item[GDK\_LINE\_DOUBLE\_DASH] Even segments are drawn, Odd segments are drawn in the background color if the fill style is \var{GDK\_SOLID}. \end{description} \var{cap\_style} determines how line ends are drawn. The following values are defined: \begin{description} \item[GDK\_CAP\_BUTT] The lines are drawn with square ends. \item[GDK\_CAP\_NOT\_LAST] Idem as \var{GDK\_CAP\_BUTT}, only for zero-width lines, the last dot is not drawn. \item[GDK\_CAP\_ROUND] The end of the line is a semicircle. The circle has diameter equal to the linewidth, and the center is the endpoint of the line. \item[GDK\_CAP\_PROJECTING] Idem as [GDK\_CAP\_BUTT], only the line extends half the linewidth outside the endpoint. \end{description} The effect of these elements will be shown in the next section. To set a color, a \var{TGDkColor} record must be allocated. Colors are specified using a RGB value. Unfortunately, not all graphics cards can show all colors. In order to find out which screen color corresponds to the RGB-specified color, the GDK uses a colormap, and allocates a color that matches the closest to the specified color values. When allocating a new color, the colormap should be specified. A colormap can be obtained from a \var{TGTKWidget} object using the \var{gtk\_widget\_get\_colormap} function; A color can then be allocated using the \var{gdk\_colormap\_alloc\_color} function: \begin{verbatim} function gdk_colormap_alloc_color(colormap:PGdkColormap; color:PGdkColor; writeable:gboolean; best_match:gboolean):gboolean; \end{verbatim} The \var{writeable} parameter specifies whether changes in the \var{color} using \var{gdk\_color\_change} are allowed. \var{best\_match} specifies whether a best match should be attempted on existing colors or an exact value is required. The function returns \var{True} if the allocation succeeded, \var{False} otherwise. \section{Drawing primitives} Using the properties introduced in the previous section, drawing can be attempted using the drawing primitives offered by GDK. GDK offers drawing functions for points, lines, segments, rectangles, polygons, circles, text and bitmaps. All functions accept as the first two parameters a \var{PGDKdrawable}, which can be a pointer to a \var{TGDKWindow} or a \var{TGDkPixmap}, and a \var{PGdkGC}, a pointer to a graphics context. These parameters are omitted from the following declarations: \begin{verbatim} procedure gdk_draw_point(x,y:gint); procedure gdk_draw_line(x1,y1,x2,y2:gint); procedure gdk_draw_rectangle(filled,x,y,width,height:gint); \end{verbatim} The above functions draw respectively a dot, a line and a rectangle. The meaning of the parameters for these functions is obvious. For the rectangle, care must be taken. If the parameter \var{Filled} is False (-1) then the drawn rectangle has actually a width and height of \var{Width+1}, \var{Height+1}. If it is filled, then the width and height are as specified in the call to \var{gdk\_draw\_rectangle}. The following functions can be used to draw a series of lines: \begin{verbatim} procedure gdk_draw_polygon(filled:gint;points:PGdkPoint; npoints:gint); procedure gdk_draw_lines(points:PGdkPoint; npoints:gint); procedure gdk_draw_segments(segs:PGdkSegment; nsegs:gint); \end{verbatim} The \var{gdk\_draw\_polygon} polygon takes a series of dots and connects them using lines, optionally filling them. The points are specified by a pointer to an array of \var{TGDKPoint} records (there should be \var{npoint} such records in the array). A \var{TGDKPoint} record contains 2 fields: \var{X,Y} which specify the location of a point. If needed, the first and last points are also connected using a line. The \var{gdk\_draw\_lines} does the same, only it cannot be filled, and it will not connect the first and last points. The \var{gdk\_draw\_segments} requires a series of \var{TGDKSegment} records. These consist of 4 fields: \var{x1,y1,x2,y2}, each describing the start and end point of a line segment. The segments will not be connected. The \var{gdk\_draw\_arc} can be used to draw a circle or a segment of the circle, or an ellipse. \begin{verbatim} procedure gdk_draw_arc(filled,x,y,width,height,angle1,angle2 : gint); \end{verbatim} The \var{x,y, width} and \var{height} parameters describe a bounding rectangle for the circle. The angles describe the start and extending angle of the segment to be drawn: The circle segment starts at angle \var{angle1} and ends at \var{angle1+angle2}. These angles are specified in 1/64ths of a degree and are measured counterclockwise, starting at the 3 o'clock direction. A circle segment drawn from 90 to 270 degrees should therefore have as angles 90*64=5760 and 270*64=17280. If filled is \var{True} (-1), then the segment will be connected to the circle centre, and filled, in effect drawing a pie-slice. Finally, for the \var{gdk\_draw\_string} function, the graphics context comes before the graphics context: \begin{verbatim} procedure gdk_draw_string(drawable:PGdkDrawable; font:PGdkFont; gc:PGdkGC; x:gint; y:gint; thestring:Pgchar); \end{verbatim} The meaning of the parameters for this functions should be obvious. The font for the \var{gdk\_draw\_string} can be obtained using the \var{gdk\_font\_load} function: \begin{verbatim} function gdk_font_load(font_name:Pgchar):PGdkFont; \end{verbatim} The font name should be specified as an X font path. All this is demonstrated in the following program: \begin{lstlisting}{} program graphics; {$mode objfpc} {$h+} uses glib,gdk,gtk,sysutils; var window, area : PGtkWidget; Function CloseApp(widget : PGtkWidget ; event : PGdkEvent; data : gpointer) : boolean; cdecl; Begin gtk_main_quit(); close_application := false; End; Function AllocateColor(R,G,B : Integer; Widget : PGtkWidget) : PGdkColor; begin Result:=New(PgdkColor); With Result^ do begin Pixel:=0; Red:=R; Blue:=B; Green:=G; end; gdk_colormap_alloc_color(gtk_widget_get_colormap(Widget), Result,true,False); end; function Exposed(Widget: PGtkWidget; event : PGdkEventExpose; Data : gpointer) : Integer; cdecl; Const Triangle : Array[1..4] of TgdkPoint = ((X:10;Y:195), (X:110;Y:195), (X:55;Y:145), (X:10;Y:195)); LineStyles : Array[1..5] of TgdkLineStyle = (GDK_LINE_SOLID, GDK_LINE_ON_OFF_DASH, GDK_LINE_DOUBLE_DASH, GDK_LINE_ON_OFF_DASH, GDK_LINE_SOLID); capstyles : Array[1..5] of TgdkCapStyle = (GDK_CAP_ROUND,GDK_CAP_NOT_LAST, GDK_CAP_BUTT, GDK_CAP_PROJECTING, GDK_CAP_NOT_LAST); Var SegTriangle : Array[1..3] of TgdkSegment; Win : pgdkWindow; gc : PgdkGC; i,seg : Integer; font : PgdkFont; Angle1,Angle2 : Longint; begin gc:=gdk_gc_new(widget^.Window); Win:=widget^.window; With Event^.area do gdk_window_clear_area (win,x,y,width,height); gdk_gc_set_foreground(gc,allocatecolor(0,0,0,Widget)); gdk_draw_rectangle(win,gc,0,5,5,590,390); gdk_gc_set_foreground(gc,allocatecolor(0,0,$ffff,Widget)); for I:=10 to 50 do gdk_draw_point(win,gc,I*10,100); gdk_gc_set_foreground(gc,allocatecolor($ffff,0,0,Widget)); for I:=10 to 50 do begin gdk_gc_set_line_attributes(gc,6,LineStyles[i div 10],CapStyles[i div 10],GDK_JOIN_MITER); gdk_draw_line(win,gc,I*10,20,I*10,90) end; gdk_gc_set_line_attributes(gc,1,GDK_LINE_SOLID,GDK_CAP_BUTT,GDK_JOIN_MITER); gdk_gc_set_foreground(gc,allocatecolor($ffff,0,$ffff,Widget)); seg:=(360 div 20) * 64; For I:=1 to 20 do gdk_draw_arc(win,gc,0,220-I*4,200-i*4,8*i,8*i,i*seg,seg*19); For I:=1 to 20 do gdk_draw_arc(win,gc,-1,380-I*4,200-i*4,8*i,8*i,(i-1)*seg,seg); gdk_gc_set_foreground(gc,allocatecolor(0,$ffff,$ffff,Widget)); gdk_draw_polygon(win,gc,0,@triangle[1],4); For I:=1 to 4 do Triangle[i].Y:=400-Triangle[i].y; gdk_draw_polygon(win,gc,-1,@triangle[1],4); gdk_gc_set_foreground(gc,allocatecolor(0,$ffff,0,Widget)); For I:=1 to 4 do Triangle[i].X:=600-Triangle[i].x; gdk_draw_lines(win,gc,@triangle[1],4); For I:=1 to 3 do begin SegTriangle[i].X1:=Triangle[i].X; SegTriangle[i].Y1:=400-Triangle[i].Y; SegTriangle[i].X2:=Triangle[i+1].X; SegTriangle[i].Y2:=400-Triangle[i+1].Y; end; gdk_draw_segments(win,gc,@segtriangle[1],3); font:=gdk_font_load('-*-helvetica-bold-r-normal--*-120-*-*-*-*-iso8859-1'); gdk_gc_set_foreground(gc,allocatecolor($ffff,$ffff,0,Widget)); For I:=1 to 4 do gdk_draw_string(win,font,gc,I*100,300,Pchar(format('String %d',[i]))); result:=0; end; Begin // Initialize GTK and create the main window gtk_init( @argc, @argv ); window := gtk_window_new( GTK_WINDOW_TOPLEVEL ); gtk_window_set_policy(PgtkWindow(Window),0,0,1); gtk_signal_connect (GTK_OBJECT (window), 'delete_event', GTK_SIGNAL_FUNC( @CloseApp ), NIL); gtk_container_set_border_width (GTK_CONTAINER (window), 10); area := gtk_drawing_area_new(); gtk_container_add( GTK_CONTAINER(window), Area); gtk_signal_connect (GTK_OBJECT (area),'expose_event', GTK_SIGNAL_FUNC(@Exposed),Nil); gtk_drawing_area_size (PGTKDRAWINGAREA(area),600,400); gtk_widget_show_all( window ); gtk_main(); end. \end{lstlisting} The main program starts by creating a main window, and adding a \var{TGTKDrawingArea} to it. It then connects 2 event handlers, one to stop the application if the window is closed (\var{CloseApp}), the other to draw the \var{TGTKDrawingArea} when it is exposed (\var{Exposed}). This latter contains the actual drawing routines, and is pretty self-explaining. It simply demonstrates the use of the drawing primitives explained above. Note that the allocated colors are not freed again, so this program does contain a memory leak. \section{Animation} The GDK drawing functions can be used to draw directly on a window visible on the screen. This is OK for normal applications, but applications that have a lot of (changing) graphics will soon see a flickering screen. Luckily, GDK provides a means to cope with this: Instead of drawing directly on the screen, one can draw on a bitmap which exists in memory, and copy parts of the bitmap to the screen on an as-need basis. This is the reason why the GDK drawing functions generally accept a \var{PGDKdrawable} parameter: This can be of the type \var{PgdkWindow} or \var{PGDKPixmap}: The \var{TGDKPixmap} can be used to do the drawing in the background, and then copy the pixmap to the actual window. This technique, known as double buffering, will be demonstrated in a small arcade game: BreakOut. The game is quite simple: at the top of the screen, there are a series of bricks. At the bottom of the screen is a small pad, which can be move left or right using the cursor keys. A ball bounces on the screen. When the ball hits a brick, the brick dissappears. When the ball hits the bottom of the window, the ball is lost. The pad can be used to prevent the ball from hitting the bottom window. When the pad hits the ball, the ball is accellerated in the direction the pad was moving at the moment of impact. Also, an idea of 'slope' is introduced: If the ball hits the pad at some distance from the pad's center, the ball's trajectory is slightly disturbed, as if the pad has a slope. After 5 balls were lost, the game is over. If all bricks have been destroyed, a next level is started. As stated above, the game will be implemented using double buffering. The ball and pad themselves will be implemented as pixmaps; the bricks will be drawn as simple rectangles. These three objects will be implemented using a series of classes: \var{TGraphicalObject}, which introduces a position and size. This class will have 2 descendents: \var{TBlock}, which will draw a block on the screen and \var{TSprite}, which contains all functionality to draw a moving pixmap on the screen. From \var{TSprite}, \var{TBall} and \var{TPad} will be derived. These two objects introduce the behaviour specific to the ball and pad in the game. The blocks will be managed by a \var{TBlockList} class, which is a descendent of the standard \var{TList} class. All these objects are managed by a \var{TBreakOut} class, which contains the game logic. The class structure could be improved a bit, but the idea is more to separate the logic of the different objects. The \var{TGraphicalObject} class is a simple object which introduces some easy access properties to get the position and size of the object: \begin{verbatim} TGraphicalObject = Class(TObject) FRect : TGdkRectangle; Public Function Contains(X,Y : Integer) : Boolean; Property Left : SmallInt Read FRect.x Write Frect.x; Property Top : SmallInt Read FRect.y Write Frect.y; Property Width : Word Read Frect.Width Write Frect.Width; Property Height : Word Read Frect.Height Write Frect.Height; end; \end{verbatim} The \var{TBlock} object is a simple descendent of the var{TGraphicalObject} class: \begin{verbatim} TBlock = Class(TGraphicalObject) Private FMaxHits : Integer; FBlockList : TBlockList; FGC : PGDKGC; FColor : PGDKColor; FNeedRedraw : Boolean; Procedure CreateGC; Function DrawingArea : PGtkWidget; Function PixMap : PgdkPixMap; Public Procedure Draw; Function Hit : Boolean; Constructor Create (ABlockList : TBlockList); Property Color : PGDKColor Read FColor Write FColor; end; \end{verbatim} The \var{FMaxHits} field determines how many times the ball must hit the brick before it dissappears. With each hit, the field is decremented by 1. The \var{FBlockList} refers to the blocklist object that will manage the block. The needed drawing widget and the pixmap on which the block must be drawn are obtained from the blockmanager using the \var{DrawingArea} and \var{Pixmap} functions. The \var{Draw} procedure will draw the block at it's position on the pixmap. The \var{Color} property determines the color in which the block will be drawn. The implementation of the \var{TBlock} methods are quite simple. The first methods don't need any explanation. \begin{verbatim} Constructor TBlock.Create (ABlockList : TBlockList); begin Inherited Create; FBlockList:=ABlockList; FMaxHits:=1; end; Function TBlock.DrawingArea : PGtkWidget; begin Result:=FBlockList.FBreakout.FDrawingArea; end; Function TBlock.PixMap : PgdkPixMap; begin Result:=FBlockList.PixMap; end; \end{verbatim} The first interesting method is the \var{CreateGC} method: \begin{verbatim} Procedure TBlock.CreateGC; begin FGC:=gdk_gc_new(DrawingArea^.Window); gdk_gc_set_foreground(FGC,FColor); gdk_gc_set_fill(FGC,GDK_SOLID); FNeedRedraw:=True; end; \end{verbatim} The method is called the first time the block must be drawn. It allocates a new graphics context using the \var{gdk\_gc\_new} function. This function accepts a pointer to a \var{TGTKWidget} as a parameter and returns a new graphics context. After the graphics context is created, the foreground color and fill style are set. (it is assumed that \var{FColor} points to a valid color) The \var{Draw} procedure actually draws the block on the pixmap, using the graphics context created in the previous method: \begin{verbatim} Procedure TBlock.Draw; begin if FGC=Nil then CreateGC; if FNeedRedraw Then begin gdk_draw_rectangle(PGDKDrawable(Pixmap),FGC,-1,Left,Top,Width,Height); FNeedRedraw:=False; end; end; \end{verbatim} The \var{FNeedRedraw} procedure is used for optimization. Finally, the \var{Hit} method is called when the block is hit by the ball. It will decrease the \var{FMaxHits} field, and if it reaches zero, the place occupied by the block is redrawn in the background color. After that, it removes itself from the blocklist and frees itself. \begin{verbatim} Function TBlock.Hit : Boolean; begin Dec(FMaxHits); Result:=FMaxHits=0; If Result then begin FBlockList.FBreakOut.DrawBackground(FRect); FBlockList.Remove(Self); Free; end; end; \end{verbatim} The \var{TSprite} object is a little more involved. The declaration is as follows: \begin{verbatim} TSprite = Class(TGraphicalObject) FPreviousTop, FPreviousLeft : Integer; FDrawingArea : PGtkWidget; FDrawPixMap : PgdkPixmap; FPixMap : PgdkPixMap; FBitMap : PGdkBitMap; Protected Procedure CreateSpriteFromData(SpriteData : PPGchar); Procedure CreatePixMap; Virtual; Abstract; Procedure SavePosition; Public Constructor Create(DrawingArea: PGtkWidget); Procedure Draw; Function GetChangeRect (Var Rect : TGDkRectAngle) : Boolean; Property PixMap : PgdkPixMap Read FPixMap; Property BitMap : PGdkBitMap Read FBitMap; end; \end{verbatim} The important property is the \var{PixMap} property; this contains the pixmap with the sprite's image. The \var{BitMap} property contains the bitmap associated with the pixmap. The second important method is the \var{GetChangeRect} method; it returns the rectangle occupied by the sprite at its previous position. This will be used to 'move' the sprite: When moving the sprite, the current position is saved (using \var{SavePosition}), and the new position is set. After that, the old position is cleared, and the sprite is drawn at the new position. All this drawing is done on the background pixmap, to avoid flickering when drawing: The result of the two drawing steps is shown at once. The implementation of the \var{Draw} method is quite straightforward: \begin{verbatim} Procedure TSprite.Draw; Var gc : PGDKGc; begin if FPixMap=Nil then CreatePixMap; gc:=gtk_widget_get_style(FDrawingArea)^.fg_gc[GTK_STATE_NORMAL]; gdk_gc_set_clip_origin(gc,Left,Top); gdk_gc_set_clip_mask(gc,FBitmap); gdk_draw_pixmap(FDrawPixMap,gc,FPixMap,0,0,Left,Top,Width,Height) gdk_gc_set_clip_mask(gc,Nil); end; \end{verbatim} After the pixmap has been created (a method which must be implemented by descendent objects), the graphics context of the drawing area is retrieved to do the drawing. The bitmap is drawn using the clipping functionality of the GDK toolkit: To this end, the clip origin is set to the position of the sprite, and the clip bitmask is set from the \var{FBitmap}, which is created when the sprite's pixmap is created. When drawing the pixmap, only the bits in the bitmap will be drawn, other bits are left untouched. The pixmap is drawn using the \var{gdk\_draw\_pixmap} function. This function copies a region from one \var{TGDKDrawable} to another. It is defined as follows: \begin{verbatim} procedure gdk_draw_pixmap(drawable:PGdkDrawable; gc:PGdkGC; src:PGdkDrawable; xsrc,ysrc,xdest,ydest,width,height:gint); \end{verbatim} The function, as all GDK drawing functions, takes a \var{PGDKDrawable} pointer and a graphics contexts as its first two arguments. The third argument is the \var{TGDKDrawable} which should be copied. The \var{xsrc,ysrc} parameters indicate the position of the region that should be copied in the source \var{TGDKDrawable}; the \var{xdest,ydest} indicate the position in the target \var{TGDKDrawable} where the bitmap should be drawn. In the case of \var{TSprite}, the function is used to copy the sprite's bitmap onto the memory pixmap with the game image. After the bitmap was copied, the clip mask is removed again. The creation of the pixmap happens when the sprite is drawn for the first time; The \var{CreateSpriteFromData} method accepts a pointer to an XPM pixmap, and uses the \var{gdk\_pixmap\_create\_from\_xpm\_d} function (explained in the previous article) to create the actual pixmap and the corresponding bitmap. \begin{verbatim} Procedure TSprite.CreateSpriteFromData(SpriteData : PPGChar); begin FPixMap:=gdk_pixmap_create_from_xpm_d(FDrawingArea^.Window, @FBitmap, Nil, SpriteData); end; \end{verbatim} This method can be used by the descendent object's \var{CreatePixmap} procedure. The \var{SavePosition} and \var{GetChangeRect} methods are very straightforward: \begin{verbatim} Function TSprite.GetChangeRect (Var Rect : TGDkRectAngle) : Boolean; begin Result:=(FPreviousLeft<>Left) or (FPreviousTop<>Top); If Result then With Rect do begin x:=FPreviousLeft; y:=FPreviousTop; Width:=Abs(Left-FPreviousLeft)+self.Width; height:=Abs(Top-FPreviousTop)+self.Height; end; end; Procedure TSprite.SavePosition; begin FPreviousLeft:=Left; FPreviousTop:=Top; end; \end{verbatim} Note that the \var{GetChangeRect} procedure returns false if the position of the sprite didn't change. This is used for optimization purposes. The pad is the simplest of the two \var{TSprite} descendents. It only adds a horizontal movement to the sprite: \begin{verbatim} TPad = Class (TSprite) Private FSlope, FSpeed,FCurrentSpeed : Integer; Protected Procedure CreatePixMap; override; Procedure InitialPosition; Public Constructor Create(DrawingArea: PGtkWidget); Procedure Step; Procedure GoLeft; Procedure GoRight; Procedure Stop; Property CurrentSpeed : Integer Read FCurrentSpeed; Property Speed : Integer Read FSpeed Write FSpeed; Property Slope : Integer Read FSlope Write FSlope; end; \end{verbatim} The procedures \var{GoLeft}, \var{GoRight} and \var{Stop} can be used to control the movement of the pad. The method \var{Step} will be called at regular intervals to actually move the pad. The \var{InitialPosition} sets the pad at its initial position on the screen. The \var{Speed} and \var{Slope} properties can be used to set the speed and slope of the pad. The implementation is quite straightforward: \begin{verbatim} \end{verbatim} \end{document}