I'm trying to create a custom scrollable text area. I created a DrawingArea and a ScrollBar inside a Grid. I have attached the draw event of DrawingArea to this.on_draw method which simply looks at ScrollBar's value and moves the Cairo.Context appropriately before drawing the Pango.Layout.
The first problem is that this.on_draw is getting invoked whenever the ScrollBar is touched even though I have not registered any events with ScrollBar. How do I prevent this, or check this?
The second problem is that even though this.on_draw is invoked, the changes made to the Context is not displayed unless the ScrollBar value is near 0 or 100 (100 is the upper value of Adjustment). Why is this happening?
I did find out that if I connect the value_changed event of ScrollBar to a method that calls queue_redraw of DrawingArea, it would invoke this.on_draw and display it properly after it. But due to the second problem, I think this.on_draw is getting invoked too many times unnecessarily. So, what is the "proper" way of accomplishing this?
using Cairo;
using Gdk;
using Gtk;
using Pango;
public class Texter : Gtk.Window {
private Gtk.DrawingArea darea;
private Gtk.Scrollbar scroll;
private string text = "Hello\nWorld!";
public Texter () {
GLib.Object (type: Gtk.WindowType.TOPLEVEL);
Gtk.Grid grid = new Gtk.Grid();
this.add (grid);
var drawing_area = new Gtk.DrawingArea ();
drawing_area.set_size_request (200, 200);
drawing_area.expand = true;
drawing_area.draw.connect (this.on_draw);
grid.attach (drawing_area, 0, 0);
var scrollbar = new Gtk.Scrollbar (Gtk.Orientation.VERTICAL,
new Gtk.Adjustment(0, 0, 100, 0, 0, 1));
grid.attach (scrollbar, 1, 0);
this.darea = drawing_area;
this.scroll = scrollbar;
this.destroy.connect (Gtk.main_quit);
}
private bool on_draw (Gtk.Widget sender, Cairo.Context ctx) {
ctx.set_source_rgb (0.9, 0.9, 0.9);
ctx.paint ();
var y_offset = this.scroll.get_value();
stdout.printf("%f\n", y_offset);
ctx.set_source_rgb (0.25, 0.25, 0.25);
ctx.move_to(0, 100 - y_offset);
var layout = Pango.cairo_create_layout(ctx);
layout.set_font_description(Pango.FontDescription.from_string("Sans 12"));
layout.set_auto_dir(false);
layout.set_text(this.text, this.text.length);
Pango.cairo_show_layout(ctx, layout);
return false;
}
static int main (string[] args) {
Gtk.init (ref args);
var window = new Texter ();
window.show_all ();
Gtk.main ();
return 0;
}
}
Also, please point out any (possibly unrelated) mistake if you find one in the above code.
The part that you are missing is that a draw signal does not mean "redraw everything". Instead, GTK+ sets the clip region of the cairo context to the part that needs to be redrawn, so everything else you do doesn't have any effect. The cairo function cairo_clip_extents() will tell you what that region is. The queue_draw_area() method on GtkWidget will allow you to explicitly mark a certain area for drawing, instead of the entire widget.
But your approach to scrollbars is wrong anyway: you're trying to build the entire infrastructure from scratch! Consider using a GtkScrolledWindow instead. This automatically takes care of all the details of scrolling for you, and will give you the overlay scrollbars I mentioned. All you need to do is set the size of the GtkDrawingArea to the size you want it to be, and GtkScrolledWindow will do the rest. The best way to do this is to subclass GtkDrawingArea and override the get_preferred_height() and/or get_preferred_width() virtual functions (being sure to set both minimum and natural sizes to the sizes you want for that particular dimension). If you ever need to change this size later, call the queue_resize() method of GtkWidget. (You probably could get away with just using set_size_request(), but what I described is the preferred way of doing this.) Doing this also gives you the advantage of not having to worry about transforming your cairo coordinates; GtkScrolledWindow does this for you.
Related
When I run the following programme:
//compile: valac --pkg gtk+-3.0 sample.vala
using Gtk;
public class Main : Object
{
private Window window;
private Gdk.Pixbuf pixbuf;
private DrawingArea da1;
public Main()
{
window = new Window();
window.destroy.connect (main_quit);
pixbuf = new Gdk.Pixbuf.from_file("sample.jpg");
var box = new Box (Orientation.VERTICAL, 5);
da1 = new DrawingArea();
da1.set_hexpand(true);
da1.set_vexpand(true);
da1.draw.connect((context) => draw_picture(context, pixbuf));
box.pack_start (da1, true, true, 0);
window.add (box);
window.show_all();
}
bool draw_picture(Cairo.Context cr, Gdk.Pixbuf pixbuf)
{
print("draw_picture\n");
int width = da1.get_allocated_width();
int height = da1.get_allocated_height();
var temp = pixbuf.scale_simple(width, height, Gdk.InterpType.BILINEAR);
Gdk.cairo_set_source_pixbuf (cr, temp, 0, 0);
cr.paint ();
return false;
}
static int main(string[] args)
{
Gtk.init(ref args);
var app = new Main();
Gtk.main();
return 0;
}
}
after start I can see 'draw_picture' printed 2 times. When I switch window to terminal, it's displayed additional 7 times. Can anyone explain why and recommend some good book explaining the details?
In order to answer that question for sure, you would need to look at how your window manager works. Probably the redraws that occur when you switch to another window are due to the widget gaining the :backdrop pseudoclass when its window ceases to be the active window.
In general a widget's draw signal is invoked whenever the window manager needs to redraw a portion of the window that includes the widget, and also whenever the widget's active style changes.
If your widget is expensive to redraw then you can avoid drawing the whole thing when it's not necessary. The Cairo context's clip region will be set to the portion which needs redrawing.
In Unity, I created a UI Scroll Rect Object. This object has two children, a Button Handler with different Buttons, and a Background.
Using it, both the Buttons and the Background are scrolling at the same speed. I want the Buttons to scroll faster than the Background, thus creating an effect of depth to the scrolling.
I can not find any options in the Objects or the Scroll Rect for this. Any ideas?
This will require a tiny bit of scripting. The best way in my opinion would involve following steps:
a) for each button, add a component that remembers its start position.
b) grab a scrollrect instance in parent
c) using it to grab an instance of ScrollBar
d) scrollbars have onValueChanged(float) callbacks, you can bind to, to know when the scroll position changes (to avoid doing checks in Update)
e) every time a scrollbar has moved (This will work regardless of whether the user used the scrollbar, or used another mean of scrolling, as the scrollrect will move the scrollbar, which will fire the event anyways), you get a callback with current scrollbar position which will be equal to normalized scrollrect position (0.1)
f) use that value to offset your localposition (something value*parallaxVector2), this should give you a nice cheap depth effect
Here's an example implementation
public class ScrolleRectDepthButton : MonoBehaviour
{
RectTransform content; // we'll grab it for size reference
RectTransform myRect;
public float parallaxAmount = 0.05f;
public Vector2 startPosition;
void Start()
{
myRect = GetComponent<RectTransform>();
startPosition = myRect.anchoredPosition;
ScrollRect scrollRect = GetComponentInParent<ScrollRect>();
content = scrollRect.content;
Scrollbar scrollbar = scrollRect.verticalScrollbar;
scrollbar.onValueChanged.AddListener(OnScrollbarMoved);
}
void OnScrollbarMoved(float f)
{
myRect.anchoredPosition = startPosition - (1 - f) * parallaxAmount * content.rect.height * Vector2.up;
}
}
I've been working on switching my applications from Swing to JavaFX. I've been working on a room escape game which displays a description of the item on which the user clicks. In Swing, I'd subclass JComponent and override the paintComponent(Graphics) method. I could draw the text there, knowing that the method is constantly called to update the screen. However, using the JavaFX Canvas, there is no method that is called constantly, which makes this task harder. I attempted save()ing the GraphicsContext after I drew the images and called restore() when I wanted to remove the text, but to no avail. Here's the important code:
package me.nrubin29.jescape;
import javafx.application.Platform;
import javafx.scene.canvas.Canvas;
import javafx.scene.canvas.GraphicsContext;
import javafx.scene.shape.Rectangle;
import java.util.Timer;
import java.util.TimerTask;
public class RoomPane extends Canvas {
private Room room;
private Toast toast;
public RoomPane() {
super(640, 480);
setOnMouseClicked(e -> {
for (JObject o : room.getObjects()) {
if (o.getBounds().contains(e.getX(), e.getY())) {
toast = new Toast(o.getDescription());
}
}
});
new Timer().schedule(new TimerTask() {
#Override
public void run() {
if (toast == null) {
return;
}
if (toast.decrement()) { // Decrements the internal counter. If the count is 0, this method returns true.
toast = null;
Platform.runLater(() -> getGraphicsContext2D().restore());
}
else {
Platform.runLater(() -> getGraphicsContext2D().strokeText(toast.getText(), 300, 100));
}
}
}, 0, 1000);
}
public void changeRoom(Room room) {
this.room = room;
GraphicsContext g = getGraphicsContext2D();
g.drawImage(room.getBackground(), 0, 0);
for (JObject o : room.getObjects()) {
g.drawImage(o.getImage(), getCenterX(o.getBounds()), getCenterY(o.getBounds()));
}
g.save();
}
}
I attempted save()ing the GraphicsContext after I drew the images and called restore() when I wanted to remove the text, but to no avail.
save and restore have nothing to with removing things like text, what they do is save in a stack the state of various settings like a stroke or fill to use to draw shapes and allow them to be popped off the stack for application later. Those routines don't effect the pixels drawn on the canvas at all.
To remove something from a GraphicsContext, you can either draw over the of it, or clear it. For your code, what you could do is snapshot the canvas node where you are trying to save it, then draw your snapshot image onto the canvas where you are trying to restore it. It is probably not the most efficient way of handling drawing (a smarter routine which just draws only damaged area where the text is would be better, but probably not required for your simple game).
However, using the JavaFX Canvas, there is no method that is called constantly
Rather than using a timer to trigger canvas calls, use a AnimationTimer or a Timeline. The AnimationTimer has a callback method which is invoked every pulse (60 times a second, or as fast as JavaFX can render frames, whichever is the lesser), so it gives you an efficient hook into the JavaFX pulse based rendering system. The Timeline can have keyframes which are invoked at user specified durations and each keyframe can have an event handler callback which is invoked at that duration.
Using the built-in JavaFX animation framework, you don't have to worry about multi-threading issues and doing things like Platform.runLater which overly complicate your code and can easily lead to subtle and serious errors.
On a kind of unrelated note, for a simple game like this, IMO you are probably better off recoding it completely to use the JavaFX scene graph rather than a canvas. That way you will be working at a higher level of abstraction rather than clip areas and repainting damaged paint components.
I am in the process of adding drag and drop support to an existing Mono/C#/GTK# application. I was wondering whether it was possible to use RGBA transparency on the icons that appear under the mouse pointer when I start dragging an object.
So far, I realized the following:
I can set the bitmap in question by calling the Gtk.Drag.SourceSetIconPixbuf() method. However, no luck with alpha transparency: pixels that are not fully opaque would get 100% transparent this way.
I also tried calling RenderPixmapAndMask() on the GdkPixbuf so that I could use Gtk.Drag.SourceSetIcon() with an RGBA colormap of my Screen. It didn't work either: whenever I started dragging, I got the following error:
[Gdk] IA__gdk_window_set_back_pixmap: assertion 'pixmap == NULL || gdk_drawable_get_depth (window) == gdk_drawable_get_depth (pixmap)' failed.
This way, the pixmap doesn't even get copied, only a white shape (presumably set by the mask argument of SetSourceIcon()) shows up on dragging.
I'd like to ask if there's a way to make these icons have alpha transparency, despite the fact that I failed to do so. In case it's impossible, answers discussing the reasons of the lack of this feature would also be helpful. Thank you.
(Compositing is - of course - enabled on my desktop (Ubuntu/10.10, Compiz/0.8.6-0ubuntu9).)
Ok, finally I solved it. You should create a new Gtk.Window of POPUP type, set its Colormap to your screen's RGBA colormap, have the background erased by Cairo to a transparent color, draw whatever you'd like on it and finally pass it on to Gtk.Drag.SetIconWidget().
Sample code (presumably you'll want to use this inside OnDragBegin, or at a point where you have a valid drag context to be passed to SetIconWidget()):
Gtk.Window window = new Gtk.Window (Gtk.WindowType.Popup);
window.Colormap = window.Screen.RgbaColormap;
window.AppPaintable = true;
window.Decorated = false;
window.Resize (/* specify width, height */);
/* The cairo context can only be created when the window is being drawn by the
* window manager, so wrap drawing code into an ExposeEvent delegate. */
window.ExposeEvent += delegate {
Context ctx = Gdk.CairoHelper.Create (window.GdkWindow);
/* Erase the background */
ctx.SetSourceRGBA (0, 0, 0, 0);
ctx.Operator = Operator.Source;
ctx.Paint ();
/* Draw whatever you'd like to here, and then clean up by calling
Dispose() on the context's target. */
(ctx.Target as IDisposable).Dispose ();
};
Gtk.Drag.SetIconWidget(drag_context, window, 10, 10);
I'm using gwt-dnd to implement drag-and-drop functionality in my GWT program. To get scrolling to work right, I need
<ScrollPanel>
<AbsolutePanel>
<VerticalPanel>
<!-- lots of draggable widgets -->
</VerticalPanel>
</AbsolutePanel>
</ScrollPanel>
I have to manually set the size of the AbsolutePanel to be large enough to contain the VerticalPanel. When I add widgets to the VerticalPanel, though, the size reported by VerticalPanel.getOffsetHeight() isn't immediately updated - I guess it has to be rendered by the browser first. So I can't immediately update the AbsolutePanel's size, and it ends up being too small. Argh!
My stop-gap solution is to set up a timer to resize the panel 500ms later. By then, getOffsetHeight will usually be returning the updated values. Is there any way to immediately preview the size change, or anything? Or, alternatively, can I force a render loop immediately so that I can get the new size without setting up a timer that's bound to be error-prone?
This is a common problem with DOM manipulations. The offsetHeight doesn't update until a short time after components are added. I like to handle this using a recursive timer until a pre-condition is violated. E.g. In your case let there be a function which adds components and will be defined as below:
public void addComponent(Widget w)
{
final int verticalPanelHeight = verticalPanel.getOffsetHeight();
verticalPanel.add(w);
final Timer t = new Timer(){
public void run()
{
if(verticalPanelHeight != verticalPanel.getOffsetHeight())
absolutePanel.setHeight(verticalPanel.getOffsetHeight() + 10 + "px");
else
this.schedule(100);
}
};
t.schedule(100);
}