MinimumScaleMode.Fit and coordinates

[krabo0om]

Hey,
thanks to @tsuijten and @peterLaurence PR #318 introduced the MinimumScaleMode.Fit mode adding whitespaces when totally zoomed out.
However, this broke the coordinates of tap events. A short example:
Lets say I have an image of 1000 width * 500 height and set MinimumScaleMode.FIT. Zoomed totally in at 1.0 and tapping the lower right I get x,y of 1000,500. Fine.
Zooming out it adds the whitespaces and centres the image, also fine. However, if I tap the lower right now I get x,y of something like 1000,2500 because the added whitespace.
This offset is calculated in ZoomPanLayout.onLayout() with translateX and translateY and should be used to offset the MotionEvent the onTouchEvent() gets.

[p-lr]

I can confirm this. I tested the changes suggested and it fixes the issue, although i'm not yet sure this is the correct way to fix this nor this won't have side effects. Nevertheless, the changes can be seen on my fork, banch fix#338.
It can now be checked with the RealMapTileView demo.

@moagrius you can check the changes, and if you agree i can PR. FYI, i logged the values of translateX and translateY during layout change. They were equal to 0 except when i zoomed out enough to see white spaces.

[moagrius]

i commented your fork peter

[moagrius]

@krabo0om this should be fixed in ad43318 - can you grab the latest master and confirm? if it works, i'll publish a release

@peterLaurence i skipped a PR (no real reason just rushing i guess) - does this look sane to you?

[p-lr]

@moagrius it looks ok for me. It fixes the markers part. One last thing is the HotSpots (aka Regions) are not offset, so it looks to me that the modified touch event position can lead to wrong HotSpot match.

[krabo0om]

@moagrius looks good to me, too.
My callout view is not chopped any more and my hot spots are all working.

[p-lr]

@krabo0om your HotSpots are working even when you have white spaces (in FIT mode)?

[krabo0om]

@peterLaurence yeah they do 😄

[p-lr]

Ok i must be missing something. Good news anyway.

[krabo0om]

there might be a new bug with callouts and clipChildren=true:

So when I show the callout I pull data from a server, the progress bar shows that. After that's done I set the visibility of the progress bar to View.GONE resulting in a new layout phase. The view itself stays visible and it's position isn't changed, however, it disappears (after the progress bar is gone) when it's not inside the MarkerLayout any longer.
I guess that the systems hides the view because it left the parent.

[p-lr]

Try to set the visibility of the marker to View.VISIBLE right after you set View.GONE on the progressbar?

[krabo0om]

the view's visibility is not affected it's always View.VISIBLE, I used a View.OnLayoutChangeListener (fired when the progress bar is gone) and a OnAttachStateChangeListener (fired when the callout is removed because of a tap on the image) to confirm this. The disappearing view behaves the same as a view inside the MarkerLayout

[krabo0om]

I could not get the CalloutLayout's moveMarker working to move it back inside the MarkerLayout to confirm that the system just ignores the callout during the render phase.

[p-lr]

I can't reproduce on my side. Callouts stay visible even when outside and when a child view's visibility is set to View.GONE.

[EDIT] I could reproduce by playing with the anchorY parameter of TileView.addCallout( View view, double x, double y, Float anchorX, Float anchorY ). I set a value so the callout goes outside the TileView. And right after this callout shows up, it disappears (no need to set a child view's vivisibility to GONE).

[p-lr]

So as a workaround, you can adjust the anchorY value for that disappearing callout, so it gets at least partially inside the visible tiles.

[krabo0om]

this works for now (I used View.offsetTopAndBottom).
I'll try to reproduce it with the demo app.

[p-lr]

I found the fix. setClipChildren( false ); has to be added in MarkerLayout constructor too.
In that particular case, the MarkerLayout was clipping its child view.

[krabo0om]

I can confirm it's working with setClipChildren( false ) in the MarkerLayout

[krabo0om]

okay I hope this is the last thing: Callouts won't be closed when tapping in the whitespace.
This should solve it (in TileView):

  @Override
  public boolean onTouchEvent(MotionEvent event) {
    mCalloutLayout.removeAllViews();
    return super.onTouchEvent( event );
  }

[moagrius]

the time difference is a killer!

anyways, IIUC i need to add 2 things:

1. setClipChildren in MarkerLayout
2. pass touch event to the CalloutLayout

i'll post those, let me know if i missed anything, otherwise i'll generate a release tomorrow

thanks guys

[moagrius]

ok it's up

[p-lr]

Looks good to me. Though i noticed that a touch inside a SampleCallout causes it to disappear, while it seems to me that it didn't before. I don't know if it's a problem anyway.

[moagrius]

it will unless the callout view consumes the event (e.g., with a click listener)

thanks peter, i'll bump versions and generate a release

i appreciate all your help, as always

[moagrius]

2.2.1 is out