Introduction

The gesture system in Flutter has two separate layers. The first layer has raw pointer events, which describe the location and movement of pointers (e.g., touches, mice, and styli) across the screen. The second layer has gestures, which describe semantic actions that consist of one or more pointer movements.

Pointers

Pointers represent raw data about the user’s interaction with the device’s screen. There are four types of pointer events:

On pointer down, the framework does a hit test on your app to determine what widgets exists at the location where the pointer contacted the screen. The pointer down event (and subsequent events for that pointer) are then dispatched to the innermost widget found by the hit test. From there, the events bubble up the tree and are dispatched to all the widgets on the path from the innermost widget to the root of the tree. There is no mechanism for canceling or stopping pointer events from being dispatched further.

To listen to pointer events directly from the widgets layer, use a Listener widget. However, generally, consider using gestures (as discussed below) instead.

Gestures

Gestures represent semantic actions (e.g., tap, drag, and scale) that are recognized from multiple individual pointer events, potentially even multiple individual pointers. Gestures can dispatch multiple events, corresponding to the lifecycle of the gesture (e.g., drag start, drag update, and drag end):

  • Tap
    • onTapDown A pointer that might cause a tap has contacted the screen at a particular location.
    • onTapUp A pointer that will trigger a tap has stopped contacting the screen at a particular location.
    • onTap A tap has occurred.
    • onTapCancel The pointer that previously triggered the onTapDown will not end up causing a tap.
  • Double tap
    • onDoubleTap The user has tapped the screen at the same location twice in quick succession.
  • Long press
    • onLongPress A pointer has remained in contact with the screen at the same location for a long period of time.
  • Vertical drag
    • onVerticalDragStart A pointer has contacted the screen and might begin to move vertically.
    • onVerticalDragUpdate A pointer that is in contact with the screen and moving vertically has moved in the vertical direction.
    • onVerticalDragEnd A pointer that was previously in contact with the screen and moving vertically is no longer in contact with the screen and was moving at a specific velocity when it stopped contacting the screen.
  • Horizontal drag
    • onHorizontalDragStart A pointer has contacted the screen and might begin to move horizontally.
    • onHorizontalDragUpdate A pointer that is in contact with the screen and moving horizontally has moved in the horizontal direction.
    • onHorizontalDragEnd A pointer that was previously in contact with the screen and moving horizontally is no longer in contact with the screen and was moving at a specific velocity when it stopped contacting the screen.

To listen to gestures from the widgets layer, use a GestureDetector.

If you’re using Material widgets, instead of using a gesture detector directly, consider using an InkWell to listen to taps, and a ListView to listen to swipes to trigger scrolling.

Gesture disambiguation

At a given location on screen, there might be multiple gesture detectors. All of these gesture detectors listen to the stream of pointer events as they flow past and attempt to recognize specific gestures. The GestureDetector widget decides which gestures to attempt to recognize based on which of its callbacks are non-null.

When there is more than one gesture recognizer for a given pointer on the screen, the framework disambiguates which gesture the user intends by having each recognizer join the gesture arena. The gesture arena determines which gesture wins using the following rules:

  • At any time, a recognizer can declare defeat and leave the arena. If there’s only one recognizer left in the arena, that recognizer is the winner.

  • At any time, a recognizer can declare victory, which causes it to win and all the remaining recognizers to lose.

For example, when disambiguating horizontal and vertical dragging, both recognizers enter the arena when they receive the pointer down event. The recognizers observe the pointer move events. If the user moves the pointer more than a certain number of logical pixels horizontally, the horizontal recognizer will declare victory and the gesture will be interpreted as a horizontal drag. Similarly, if the user moves more than a certain number of logical pixels vertically, the vertical recognizer will declare victory.

The gesture arena is beneficial when there is only a horizontal (or vertical) drag recognizer. In that case, there will be only one recognizer in the arena and the horizontal drag will be recognized immediately, which means the first pixel of horizontal movement can be treated as a drag and the user will not need to wait for further gesture disambiguation.