Part of my particular dilemma is that I would like to be able to get the initial position of a drag gesture. Depending on the initial position of that drag gesture, the application would either 1) pan the view or 2) display a menu, but not perform these gestures at the same time (which is where part of struggle lies).
For example, if the user initiated a drag from the very left side of their screen and dragged inwards, a menu would pop in instead of the view panning.
I would also like to be able to execute a double tap gesture without also activating a tap gesture, if that's at all possible. I've tried working with boolean flags - for example,
// ...
if (gesture.GestureType == GestureType.DoubleTap)
{
isDoubleTap == true;
}
// ...
public static Vector2 Tap
{
get
{
if (!isDoubleTap)
return gesture.Position;
// ...
But that doesn't work.
I've tried using TouchCollection - if anyone would like me to elaborate on my issues with that I can, but for now I'll say what I tried hasn't worked. It's entirely possible I may have just goofed as I am a novice when it comes to working with touch input.
I've been working on this for a few days and have done as much searching as I can, and nothing I've found has alleviated my issue - if I happened to have missed something, I apologize.
Thank you for your time!
Concerning start position of a drag:
There is a gesture for a drag ending, so if you receive a drag and its the first one since the last drag ended, thats the initial position.
Concerning tap/doubletap:
MonoGame works the same way as XNA as documented here:
The user tapped the screen twice in quick succession. This
always is preceded by a Tap gesture.
This sounds like a input-bindings design problem more than a technical question imo. Consider also what you can move to instead occur on press or release rather than only making use of gestures.
Related
I need the application idle time in my software. For that reason, I made a helper class ApplicationIdleHelper which implements the IMessageFilter interface.
This works fine and if my application is in idle for some time, I'm showing a DevExpress WaitForm using this line of code:
SplashScreenManager.ShowForm(typeof(WaitForm));
In this WaitForm I show the user some information about what's being done in the background. If the user moves the mouse or presses some keys I close the WaitForm like this:
SplashScreenManager.CloseForm();
Here's the problem explained in steps:
Mouse cursor is on the form.
User doesn't do anything for some time -> idle time -> so I show the WaitForm.
Now I get a MouseMove message in my PreFilterMessage method? BUT WHY? Mouse didn't move. No keys pressed? Because I get a MouseMove message my application thinks, the user did some input and automatically closes the WaitForm.
Same behavior if I close the WaitForm.
Here's a sample application, so you should be able to reproduce the behavior:
https://drive.google.com/file/d/0BxabrokJG-OWV3FLV2hNNVk5NjQ/view?usp=sharing
The DevExpress documentation says:
Wait Forms and Splash Screens are displayed by a Splash Screen Manager
in a separate thread.
Maybe this has something to do with that behavior?
Hope somebody can explain me, why I geht a MouseMove message in my PreFilterMessage function, after showing or closing the WaitForm.
Thank you in advance.
The most likely cause of this is that the mouse is sensitive to environmental noise. It's entirely possible for a mouse to experience a little bit of jitter that causes it to report very small movements, which ultimately work out to zero change in position. Alternatively, and this isn't verified, Windows or some other software on the system could be generating extra mouse move messages to make sure that everyone stays in sync with the current mouse position.
Either way, the most stable solution is to decide on an amount of motion you consider "real" (see threshold below), and then:
Capture the mouse position when you're going to sleep.
Every time you get a WM_MOUSEMOVE message (or a MouseMove event) calculate the amount of that motion, as in:
Point cached; // from when you went to sleep
Point current; // determined from the window message/event
double move = Math.Sqrt(Math.Pow(cached.X - current.X, 2) +
Math.Pow(cached.Y - current.Y, 2))
if (move > threshold)
{
// Wake up
}
else
{
// Ignore and optionally update the cached position
// in case the mouse is slowly drifting
}
(Note that you don't necessarily need to calculate the real distance that way, you could just use ΔX+ΔY)
Whenever you're dealing with hardware, you need to be ready for it to send you updates that you aren't expecting. Pressing a button for example, can cause the physical contact to bounce, which causes multiple press/break signals at the electrical level. Most of the time, the hardware is designed to filter the noise, but sometimes this seeps through.
I am implementing a custom drag and drop interface with winForm Buttons and after viewing several solutions on how to obtain mouse position and check it against control bound have not been able to get it to work.
I have tried:
button.ClientRectangle.Contains(PointToClient(Cursor.Position))
and
button.ClientRectangle.Contains(PointToClient(Control.MousePosition))
Both of these have failed to work. Checking the mouse bounds seem like a simple operation, but I really am stumped.
My speculation of the unexpected values are:
Process of obtaining cursor position may be in wrong corner of cursor image
Method/Function does not work on Buttons for some reason
You are using the wrong object reference, calculating the mouse position relative to the form instead of the button. And you are writing it in a way that make this very hard to debug. Fix:
var pos = button.PointToClient(Cursor.Position);
System.Diagnostics.Debug.WriteLine(pos); // Now it is easy
if (button.ClientRectangle.Contains(pos)) {
// etc...
}
I've implemented "ChangeCase" keyboard shortcut (like Shift+F3 in MS WORD) for RichTextBox, which changes the text either selected by mouse, or the last word before caret's position. The problem is, it SOMETIMES loses the selection, or moves the caret one word left.
Once it changes the textcase without this changing of caret position, then it never changes the caret position (propably some WPF's internal caching.), so it can only happen the first time i run this function to a portion of text.
The code used is the solution mentioned in here WPF Flowdocument "change case" feature .
One problematic section of code is certainly
end = this.CaretPosition;
EditingCommands.MoveLeftByWord.Execute(null, this);
start = this.CaretPosition;
this.CaretPosition = end;
However I have no idea why it only occurs sometimes and how to fix this.
I thing it has something to do with the execution speed of this Execute() method and some side effects, because at my WPF app it only happens sometimes, but when hosting this WPF control in Winforms, moving the caret one word left happens all the time (if I hold Shift+F3, the cursor moves word by word to the very beginning of the document)
Other problem can be related with changing text of a TextRange, which is resulting in losing the selection? But again, it doesnt happen all the time and I have no clue how to fix it.
Any ideas?
I ended up with 2 options, ignoring this error or implementing the
MoveLeftByWord
logic manully without touching the
CaretPosition
I came across the weirdest situation while code-generating polygons and attaching them to a Virtual Earth 3D Globe Control. I have enabled mouse controls, as discussed on this thread in codeplex: http://bingmapswpf.codeplex.com/discussions/279548
Context: Map with several polygonGeometries, some of them are intentionally overlayed (using z-index).
*strong text*Actions: Click over a given polygonGeometry or trigger MouseEnter/Leave event over a given polygonGeometry.
Result: The object isn't detected by the click or by the MouseEnter/Leave event, however if I apply some "offset" to my clicks/hovers over the PolygonGeometry, the events pickup the object on a "empty space" in the map, a couple of pixels away for the actual object.
Aditional info: This behaviour goes away completely if I zoom in the object (almost to a full screen size), and starts getting worse as I zoom out. In High-level views of the map/objects it's impossible to click or hover any objects, or at least they don't get picked up by the events.
So, right now my theory is that by some reason, in lower zoom scenarios, the map "missplaces" the polygonGeometrys (although it seems that they are drawn properly). Being the object missplaced, no shapeId/layerId are detected and thus no action over the event is triggered.
So, I would like to know if someone has already came across this situation and how it was fixed, and/or if I'm doing something wrong on my development (check the mouse events adventure on the post mentioned in the beginning of this discussion), because this is on annoying problem that just doesn't go away... Any suggestion, tip or theory is most welcome!
Thanks in advance for reading and helping. Sorry for any bad english,
-RG
I have a Pivot which contains a WebBrowser control that practically takes up the whole page (appart from the Pivot header of course).
I would like to figure out how to make the WebBrowser control allow for the user to swipe left/right to activate the Pivot control. Currently it just pans the WebBrowser control left/right
Can this be done??
Thank
While I cannot tell you exactly how to pass the swipes to the pivot, I can tell you how to do a part of the job: how to catch/analyze/disable custom gestures over the WebBrowser.
If I remember correctly, in the 7.0:
the WebBrowser component consisted almost only of an internal TileHost wrapped in some grids/borders
the TileHost did all the work related to processing touch events
the TileHost did it completely internally (in the native layer), without the .Net seeing any manipulation-events (I think), or at least it ignored all the attempts to handle/override the manipulation-event on the upper layer. The WebBrowserInterop class was mostly empty in these matters.
Now, in the 7.5 that I have (maybe on 7.1 too, I dont know), it seems that the MS is working really hard on some WebBrowser manipulation problems --- I think they are working towards having the scrolling/swiping fully processed by the .Net layer. They have written a special class PanZoomContainer and injected them into the VisualTree of WebBrowser's internal template. The WebBrowserInterop was greatly enriched with many tunnels for event notifications. The WebBrowserInterop hooks into PanZoomContainer's ManipulationEvents, then passes them to the native layer. Also, it listens to events/commands from the native layer, called for example "ZoomAndScroll" or "ShowSIP" - and mostly passes them back to the PanZoomContainer. The idea is crystal clear right? They have rewired the event handling from completely-internal to a bit of spaghetti, but have achieved passing them through the PanZoomC.
Now, whats in that for me/us?
It is the PanZoomContainer, whose Mani-Events are inspected. The TileHost does not capture them now. In this version of the WebBrowser control, it's VisualTree consists of some borders, grids, a PanZoomContainer and a TileHost (the renderer). It is similar to that:
WebBrowser
PanZoom
ContentPresenter
Border/Name="border" <- you want this one
TileHost
I've skipped a few Borders and Grids, they are mostly irrelevant to the problem. Now, if the PanZoomContainer's Mani-Events are listened to, let's block them!
Using VisualTreeHelper, just dig deeper and deeper until you find a FrameworkElement.Name=="border". This is the border that wraps the TileHost that is the "renderer" that takes 99% space of the control. Be warned that there's a ContentPresenter, so you may have to wait until the controltemplate gets instantiated (ie. Loaded/LayoutUpdated).
Once you have your hands on that Border, attach all Mani-Event handlers to it: started, delta and completed. PanZoom is a normal xaml/silverlight/.net/etc control, so it actually obeys e.Handled = true :) Now, if you want to disable ie. vertical scrolling, filter the Delta and Completed events for Translation.Y<>0. If you want to disable tapping but leave srolling/panning - filter X==0&Y==0.
And that was the easy part.
The hard part is to experiment with filtering on different Start/Delta/Stop and adjusting the behaviour to your likes.
Although it might look very nice and tempting, this will NOT get you any real/nice results easily. For example, I wrote "if you want to disable vertical scrolling, then set a filter 'if y==0 then e.handled=true' ". Great? easy? Not!
Assume we want to "disable bouncy horizontal panning" while leaving "vertical scrolling". or vice versa, whatever, it is only an example:
Here's a small rectangular device with a sensitive touchscreen. Please make such a vertical swipe/pan/drag on the screen, that the resulting X-compound will be ZERO. If you set such filter, it will be almost impossible to it properly. Your users will want to kill you for forcing them to retry-that-vertical-scrolling for five or more times, until they make a perfect vertical swipe.
Of course you can make it not ==0, but leave some small margin. ok. But if you make the margin too big, the control will catch the intermediate offaxis movement and make a tiny horizontal pan also.. After a few unlucky vertical swipes, the total horizontal pan may accumulate from those small leftovers will accumulate and the diplacement maybe will be noticeable.
But there's some another vile side effects:
Saying shortly, you have commited e.Handled=true. The event is GONE. Dead. Deased. if you just wanted the WebBrowser to SKIP for example horizontal swipes, so that the outer (Pivot) control notices them and processes..... whoops. The event is GONE. Earlier, the TileHost/PanZoomC have extinguished the events, now you have it done yourself. Sounds like a bad joke, eh?
Fortunatelly:
since you have attached your handlers to the bottommost "border", they may not only block the events, but may also actually listen&publish them elsewhere. That is, if those handlers detect an interesting movement, they may e.Handled=true on it, but at the same time they can notify your custom objects about that discovery, and ie. start your storyboards.
mani-events are at hand, but there is also a second layer that listens to the manipulations: the GestureListener/GestureService from the Silverlight Toolkit. It reports events after they are handled by mani-events, but it reports them with no regard to any e.Handled=true that were set on them. It is completely separate gesture-listening mechanism, and you can also use it to detect manipulations that were 'cancelled'
.. and so the fun goes like that and maybe even a little further.
This is similar to putting a Map inside a Pivot - which is discussed here - http://mine.tuxfamily.org/?p=111 - already mentioned in quite a few questions - https://stackoverflow.com/search?q=mine.tuxfamily.org
In general, the advice seems to be usability based:
try to avoid putting controls which use Touch inside the PivotItem's
As an aside, if you are just using the web browser control for a very small amount of static html (so you don't need any scrolling at all) then you could just remove HitTesting from the web browser.
I do not know WP7 Pivot, but are there any Preview* events on the Pivot control that allow you to handle the touches and mark them as processed?
Call the below method and pass your parameter as PivotControl x:name and WebBrowserControl x:name to this method.
Here the WebBrowserControl is placed in second pivot item i.e. Pivot Index is 1 and I am trying to swipe left or right and reach to pivot index 2 or 1 respectively.
public static void SwipteLeftRight(Microsoft.Phone.Controls.Pivot pivotControl, Microsoft.Phone.Controls.WebBrowser webBrowserControl)
{
var gesListener = GestureService.GetGestureListener(webBrowserControl);
gesListener.Flick += ((sen, args) =>
{
if (args.Direction == System.Windows.Controls.Orientation.Horizontal)
{
if (args.HorizontalVelocity < 0)
{
if (((Microsoft.Phone.Controls.PivotItem)(pivotControl.SelectedItem)).Header.ToString().Trim() == "Pivot Item name")
{
pivotControl.SelectedIndex = 2; //Next Pivot item
}
}
else if (args.HorizontalVelocity > 0)
{
if ((Microsoft.Phone.Controls.PivotItem)(pivotControl.SelectedItem)).Header.ToString().Trim() == "Pivot Item name")
{
pivotControl.SelectedIndex = 0; // Previous Pivot Item
}
}
}
});
}
It worked for me. Cheers