assorted_layout_widgets #

I will slowly but surely add interesting widgets, classes and methods to this package.

Despite the package name, they are not only related to layout. Here they are:

ColumnSuper
RowSuper
FitHorizontally
Box
WrapSuper
ButtonBarSuper
TextOneLine
Delayed
Pad
NormalizedOverflowBox
showDialogSuper and showCupertinoDialogSuper
TimeBuilder
GlobalValueKey and GlobalStringKey
MaskFunctionTextInputFormatter
SideBySide
Button and CircleButton
CaptureGestures
NonUniformOutlineInputBorder
NonUniformRoundedRectangleBorder

Note: The widgets you don't use will be removed by Flutter's tree shaking. So feel free to add the library even if you want to use only one of them.

ColumnSuper #

Given a list of children widgets, this will arrange them in a column. It can overlap cells, add separators and more.

ColumnSuper({
  List<Widget> children,
  double outerDistance,
  double innerDistance,
  bool invert,
  Alignment alignment,
  Widget separator,
  bool separatorOnTop,  
  bool removeChildrenWithNoHeight,
});

children is the list of widgets that represent the column cells, just like in a regular Column widget. However, the list may contain nulls, which will be ignored.
outerDistance is the distance in pixels before the first and after the last widget. It can be negative, in which case the cells will overflow the column (without any overflow warnings).
innerDistance is the distance in pixels between the cells. It can be negative, in which case the cells will overlap.
invert if true will paint the cells that come later on top of the ones that came before. This is specially useful when cells overlap (negative innerDistance).
alignment will align the cells horizontally if they are smaller than the available horizontal space.
separator is a widget which will be painted between each cell. Its height doesn't matter, since the distance between cells is given by innerDistance (in other words, separators don't occupy space). The separator may overflow if its width is larger than the column's width.
separatorOnTop if true (the default) will paint the separator on top of the cells. If false will paint the separator below the cells.
removeChildrenWithNoHeight if true, children with zero height will not result in an extra innerDistance and separator. If all children have zero height, the outerDistance will also be removed. In other words, it's as if children with zero height are removed, except for the fact they still occupy width. The default is false. See this interactive example.

Note: This is not a substitute for Flutter's native Column, it doesn't try to have a similar API, and it doesn't do all that Column does. In special, Expanded and Flexible widgets don't work inside of ColumnSuper, and it will overflow if the column is not big enough to fit its contents. ColumnSuper is meant only for certain use cases where Column won't work, like when you need overlapping cells or separators.

Try running the ColumnSuper example.

Also, try ColumnSuper Playground.

RowSuper #

Given a list of children widgets, this will arrange them in a row. It can overlap cells, add separators and more.

RowSuper({
  List<Widget> children,
  double outerDistance,
  double innerDistance,
  bool invert,
  Alignment alignment,
  Widget separator,
  bool separatorOnTop,
  bool fitHorizontally,
  double shrinkLimit,
  MainAxisSize mainAxisSize,
});

On contrary to ColumnSuper and the native Row (which will overflow if the children are too large to fit the available free space), RowSuper may resize its children proportionately to their minimum intrinsic width.

Try running the RowSuper example.

Also, try RowSuper Playground.

Most parameters are the same as the ones of ColumnSuper, except:

fill if true will force the children to grow their widths proportionately to their minimum intrinsic width, so that they fill the whole row width. This parameter is only useful if the children are not wide enough to fill the whole row width. In case the children are larger than the row width, they will always shrink proportionately to their minimum intrinsic width, and the fill parameter will be ignored. See: RowSuper Fill example.
fitHorizontally if true will shrink the children, horizontally only, until the shrinkLimit is reached. This parameter is only useful if the children are too wide to fit the row width. row width. Avoid using fitHorizontally together with fill: true.
shrinkLimit by default is 67%, which means the cell contents will shrink until 67% of their original width, and then overflow. Make shrinkLimit equal to 0.0 if you want the cell contents to shrink with no limits. Note, if fitHorizontally is false, the shrinkLimit is not used.
mainAxisSize by default is MainAxisSize.min, which means the row will occupy no more than its content's width. Make it MainAxisSize.max to expand the row to occupy the whole horizontal space.

You can also use a RowSpacer to add empty space (if available) between cells. For example:

RowSuper(
  children: [
    widget1,
    RowSpacer(),
    widget2,
    widget3,
    ],
);

Try running the RowSuper with FitHorizontally example.

Note: This is not a substitute for Flutter's native Row, it doesn't try to have a similar API, and it doesn't do all that Row does. In special, Expanded and Flexible widgets don't work inside of RowSuper, since RowSuper will resize cells proportionately when content doesn't fit. RowSuper is meant only for certain use cases where Row won't work, like when you need overlapping cells, or when you need to scale the contents of the cells when they don't fit.

FitHorizontally #

FitHorizontally({
  Widget child,
  double shrinkLimit,
  bool fitsHeight,
  AlignmentGeometry alignment,
});

The child will be asked to define its own intrinsic height. If fitsHeight is true, the child will be proportionately resized (keeping its aspect ratio) to fit the available height.

Then, if the child doesn't fit the width, it will be shrinked horizontally only (not keeping its aspect ratio) until it fits, unless shrinkLimit is larger than zero, in which case it will shrink only until that limit. Note if shrinkLimit is 1.0 the child will not shrink at all. The default is 0.67 (67%).

This is specially useful for text that is displayed in a single line. When text doesn't fit the container it will shrink only horizontally, until it reaches the shrink limit. From that point on it will clip, display ellipsis or fade, according to the text's Text.overflow property.

Note: FitHorizontally with shrinkLimit 0.0 is not the same as FittedBox with BoxFit.fitWidth, because FitHorizontally will only scale horizontally, while FittedBox will maintain the aspect ratio.

Try running the FitHorizontally example.

Box #

Box is something between a Container and a SizedBox, which is less verbose and can be made const.

const Box({
  bool show = true,
  Color color,
  EdgeInsetsGeometry padding,
  double width,
  double height,
  Alignment alignment,
  Widget child,
});

Since Box can be made const, it's good for creating colored boxes, with or without a child and padding:

const Box(color: Colors.red, width: 50, height:30);

It allows you to make const large blocks of code. For example, the following code uses a Box, and couldn't be const if we were to use a SizedBox or a Container:

static const progressIndicator =
  Opacity(
    opacity: 0.6,
    child: Box( // Can't use SizedBox or Container here. 
      color: Colors.blue,
      alignment: Alignment.center,
      child: Padding(
          padding: Pad(all: 5.0),
          child: AspectRatio(
              aspectRatio: 1,
              child: CircularProgressIndicator(
                  valueColor: AlwaysStoppedAnimation<Color>(Colors.white)))),),);}

Const objects are final/immutable and created in compile time. So you don't waste time creating them. Also, all const objects of the same type with the same parameters are the same instance. So you don't waste memory creating more than one of them. In other words, const objects make your program faster and more memory efficient. They can also be used as default values in constructors, and they work well with hot-reload, while final values do not.

Extra Box features #

You can hide the box by making the show parameter equal to false.
If you make removePaddingWhenNoChild: true, the padding is only applied if the child is not null. If the child is null and width and height are also null, this means the box will occupy no space (will be hidden).
Note: You can use the Pad class (provided in this package) for the padding, instead of EdgeInsets. For example:
```
Box(padding: Pad(horizontal: 8, top: 20));
```
You can change a box with the copyWith method. For example:
```
myBox.copyWith(color: Colors.blue);
```

You can create boxes by adding a Box to one these types: bool, Color, EdgeInsetsGeometry, AlignmentGeometry, or Widget:

// To hide the box:
Box(...) + false;
  
// To show the box:
Box(...) + true;
  
// To change the box color:
Box(...) + Colors.green;
  
// To change the box padding:
Box(...) + Pad(all: 10);
  
// To substitute the box child:
Box(...) + Text('abc');
  
// To put a box inside of another:
Box(...) + Box(...);

Note: If you add null, that's not an error. It will simply return the same Box. However, if you add an invalid type it will throw an error in RUNTIME.

// Not an error:
Box(...) + null;
  
// Throws:
Box(...) + MyObj();

Methods to change width and height of the box: add, subtract.

Debugging: #

If you need to quickly and temporarily add a color to your box so that you can see it, you can use the constructors Box.r for red, Box.g for green, Box.b for blue, and Box.y for yellow.
```
Box(child: myChild);
Box.r(child: myChild);
Box.g(child: myChild);
Box.b(child: myChild);
Box.y(child: myChild);
```
If you want to see rebuilds, you can use the Box.rand constructor. It will then change its color to a random one, whenever its build method is called.
```
Box.rand(child: myChild);  
```

All these debugging constructors are marked as deprecated so that you don't forget to remove them.

WrapSuper #

WrapSuper is similar to the native Wrap widget with direction = Axis.horizontal, but it allows you to choose different algorithms for the line-breaks.

WrapSuper displays its children in lines. It will leave spacing horizontal space between each child, and it will leave lineSpacing vertical space between each line. The contents of each line will then be aligned according to the alignment. The algorithm for the line-breaks is chosen by wrapType.

WrapSuper({
  Key key,
  WrapType wrapType,
  double spacing,
  double lineSpacing,
  WrapSuperAlignment alignment,
  List<Widget> children,
});

WrapSuper with WrapType.fit uses a greedy algorithm for line breaks, which is the same one used by the native Wrap widget.

However, WrapSuper with WrapType.balanced (the default) uses a minimum raggedness algorithm for line breaks. It will position its child widgets in the same number of lines as the greedy algorithm, but these lines will tend to be more similar in width.

For example:

Here is my original StackOverflow question that resulted in this widget.
The algorithm I used was based on this one (Divide and Conquer), which always considers spacing: 1.0. It was changed (with the help of CodeChef) to allow for other spacings.
Add your thumbs up here if you want native Text widgets to also allow for better line-breaks.

WrapFit #

After WrapSuper distributes its children in each line, the wrapFit parameter defines the width of the widgets:

min (the default) will keep each widget's original width.
divided will make widgets fit all the available horizontal space. All widgets in a line will have the same width, even if it makes them smaller that their original width.
proportional will make widgets larger, so that they fit all the available space. Each widget width will be proportional to their original width.
larger will make widgets larger, so that they fit all the available space. Will try to make all widgets the same width, but won't make any widgets smaller than their original width. In more detail: 1) First, divide the available line width by the number of widgets in the line. That is the preferred width. 2) Keep the width of all widgets larger than that preferred width. 3) Calculate the remaining width and divide it equally by the remaining widgets.

Some examples:

Try running the WrapFit example.

ButtonBarSuper #

ButtonBarSuper has a similar API to a regular ButtonBar, but will distribute its buttons by using a WrapSuper.

The default (which may be changed) is WrapType.balanced and WrapFit.larger, which means it will distribute the buttons in as little lines as possible in a balanced way; will make the buttons fill all the available horizontal space; and will try to make buttons have similar width in each line, without reducing their widths.

Try running the ButtonBarSuper example.

TextOneLine #

TextOneLine is a substitute for Text when maxLines: 1, to fix this issue: https://github.com/flutter/flutter/issues/18761 filled by myself a long time ago.

It renders ellipsis as expected, much better than the current/ buggy and ugly-looking ellipsis of the native Text widget, which cuts the whole word.

For example, this:

Text("This isAVeryLongWordToDemonstrateAProblem", maxLines: 1, softWrap: false);

Will print this in the screen:

This ...

While this:

TextOneLine("This isAVeryLongWordToDemonstrateAProblem");

Will print this:

This isAVeryLongWordToDemonst...

Delayed #

Delayed can be used to give a widget some initial value, and then, after some delay, change it to another value. As we'll see, Delayed is specially useful when used with implicitly animated widgets.

As soon as Delayed is inserted into the tree, it will build the widget returned by builder with initialized==false. Then:

If delay is null, it will rebuild with initialized==true in the next frame (usually 16 milliseconds later).
If delay is NOT null, it will rebuild with initialized==true after that delay.

For example, this shows a widget after a 2 seconds delay:

Delayed(delay: const Duration(seconds: 1),
  builder: (context, bool initialized) =>
    initialized
      ? Container(color: Colors.red, width: 50, height: 50)
      : SizedBox()));

For example, this changes a widget color after a 3 seconds delay:

Delayed(delay: const Duration(seconds: 3),
  builder: (context, bool initialized) =>
    Container(color: initialized ? Colors.red : Colors.blue,
              width: 50, 
              height: 50)
    )
)

For example, this will fade-in a widget as soon as it enters the screen:

Delayed(
  builder: (context, bool initialized) =>
    AnimatedOpacity(opacity: initialized ? 1.0 : 0.0,
                    duration: const Duration(seconds: 1),
                    child: MyWidget()
    )
);

For example, this will fade-in a widget 300 milliseconds after it enters the screen:

Delayed(delay: const Duration(milliseconds: 300),
  builder: (context, bool initialized) =>
    AnimatedOpacity(opacity: initialized ? 1.0 : 0.0,
                    duration: const Duration(seconds: 1),
                    child: MyWidget()
    )
);

Try running the Delayed example.

Pad #

Pad is an EdgeInsetsGeometry which is easy to type and remember.

For example, instead of writing padding: EdgeInsets.symmetric(vertical: 12) you can write simply padding: Pad(vertical: 12).

// Instead of EdgeInsets.all(12)
padding: Pad(all: 12)

// Instead of EdgeInsets.only(top: 8, bottom: 8, left: 4, right: 2)
padding: Pad(top: 8, bottom: 8, left: 4, right: 2)

// Instead of EdgeInsets.symmetric(vertical: 12)
padding: Pad(vertical: 12)

// Instead of EdgeInsets.symmetric(vertical: 12, horizontal: 6)
padding: Pad(vertical: 12, horizontal: 6)

You can also compose paddings. For example, if you want 40 pixels of padding in all directions, except the top with 50 pixels: padding: Pad(all: 40, top: 10).

During development, you sometimes need to temporarily remove the padding, for debugging reasons. Unfortunately you can't just comment the padding parameter, because the Padding widget doesn't accept null padding. But you can just add .x to the Pad class to remove it. It's marked as deprecated so that you don't forget to change it back to normal:

// This is the same as Pad.zero.
padding: Pad.x(top: 8, bottom: 8, left: 4)

NormalizedOverflowBox #

A NormalizedOverflowBox is a widget that imposes different constraints on its child than it gets from its parent, possibly allowing the child to overflow the parent.

A NormalizedOverflowBox is similar to an OverflowBox. However, then OverflowBox may throw errors if it gets constraints which are incompatible with its own constraints. For example, if an OverflowBox is inside a container with maxWidth 100, and its own minWidth is 150, it will throw:

The following assertion was thrown during performLayout():
BoxConstraints has non-normalized width constraints.

The NormalizedOverflowBox, on the other hand, will just make sure maxWidth is also 150 in the above example, and throw no errors. In other words, a NormalizedOverflowBox is safer to use, and in my opinion has the behavior OverflowBox should have had.

Try running the NormalizedOverflowBox Example. Then substitute the NormalizedOverflowBoxs with regular OverflowBoxes and see where it fails.

showDialogSuper and showCupertinoDialogSuper #

Functions showDialogSuper and showCupertinoDialogSuper are identical to the native showDialog and showCupertinoDialog, except that they let you define an onDismissed callback for when the dialog is dismissed:

showDialogSuper(
   context: context,
   onDismissed: (dynamic result) { print("Dialog dismissed"); }
   builder: ...
}

Usually there are 3 ways to close a dialog:

Pressing some button on the dialog that closes it (usually by calling Navigator.pop(context)).
Tapping the barrier.
Pressing the Android back button.

All three ways will result in the onDismissed callback being called.

However, when the dialog is popped by Navigator.of(context).pop(result) you will get the result in the onDismissed callback. That way you can differentiate between the dialog being dismissed by an Ok or a Cancel button. The result is null when the dialog is dismissed by tapping the barrier or pressing BACK in Android. Example:

showDialogSuper<int>(
...
  actions: [
    ElevatedButton( onPressed: (){Navigator.pop(context, 1);}, child: const Text("OK"),
    ElevatedButton( onPressed: (){Navigator.pop(context, 2);}, child: const Text("CANCEL"),        
  ]
  ...
  onDismissed: (int? result) {
    if (result == 1) print("Pressed the OK button.");
    else if (result == 2) print("Pressed the CANCEL button.");
    else if (result == null) print("Dismissed with BACK or tapping the barrier.");  
  });

This method was created to solve this issue: https://github.com/flutter/flutter/issues/26542 filled by myself a long time ago.

Try running the showDialogSuper Example.

TimeBuilder #

If you need some widget to change periodically (clocks, countdowns, stopwatches etc.), one way of implementing this is using a Timer to rebuild it. This is, however, very inefficient and may make your app slow.

The TimeBuilder widget gives you the correct implementation of periodic rebuilds. It's based on Remi Rousselet's code and you can read all about it here.

Apart from better performance, using the TimeBuilder widget has the following advantages:

Compatible with DevTools "slow animations", which reduce the speed of AnimationControllers.
Compatible with Clock changes, allowing for testing (skip frames to target a specific moment in time).
The TimeBuilder widget animation is "muted" when the widget is not visible. For example, when the widget is a route that is currently not visible, or because of an ancestor widget such as Visibility.

Let's see some examples.

Periodical #

To create a clock that ticks once per second, you can use the .eachSecond constructor:

TimeBuilder.eachSecond(
   builder: (BuildContext context, DateTime now, int ticks, bool isFinished) 
                => MyClock(now),
);

If you want the rebuilds to stop after 30 seconds you can add the seconds parameter:

TimeBuilder.eachSecond(
   seconds: 30,
   builder: (BuildContext context, DateTime now, int ticks, bool isFinished) 
                => MyClock(now),
);

There are also .eachMillisecond, .eachMinute and .eachHour constructors.

Countdown #

You can also create a seconds countdown from a certain DateTime:

// 100 seconds countdown.
TimeBuilder.countdown(
   start: DateTime.now(),
   seconds: 15,
   builder: (BuildContext context, DateTime now, int ticks, bool isFinished,
      {required int countdown}) 
         => Text(isFinished ? "FINISHED" : countdown.toString()),
);

General animation #

You can also create a general animation with the .animate constructor:

TimeBuilder.animate(
   builder: (BuildContext context, DateTime now, int ticks, bool isFinished) 
                => MyWidget(now),
   isFinished: ({required DateTime currentTime, required DateTime lastTime, required int ticks,})
                => currentTime.difference(initialTime) > Duration(seconds: 100),               
);

Creating your own #

And, finally, you can also create your own TimeBuilder using the default constructor:

const TimeBuilder({
   Key? key,
   required this.builder,
   required this.ifRebuilds,
   this.isFinished,
}) : super(key: key);

You must provide the builder and ifRebuilds callbacks.

For each frame, Flutter will first call your ifRebuilds callback, which may return true or false:

typedef IfRebuilds = bool Function({

  /// The current time.
  required DateTime currentTime,

  /// The time of the last tick.
  required DateTime lastTime,

  /// The number of ticks since the timer started.
  required int ticks,
});

Only when it returns true, the builder will be asked to generate a widget. The builder callback is of type TimerWidgetBuilder:

typedef TimerWidgetBuilder = Widget Function(
  BuildContext context,

  /// The time of the current tick.
  DateTime dateTime,

  /// The number of ticks since the timer started.
  int ticks,

  /// This is false while the timer is on, and becomes true as soon as it ends.
  bool isFinished,
);

There is also an optional isFinished callback. Returning true here will generate one last rebuild, and then stop the animation for good (no more rebuilds).

GlobalValueKey and GlobalStringKey #

For local keys, Flutter provides ObjectKey and ValueKey. But for global keys, it provides only GlobalObjectKey, which compares by identity:

Key keyA = GlobalObjectKey('1' + '23');
Key keyB = GlobalObjectKey('12' + '3');

keyA == keyA; // true
keyA == keyB; // false

This package provides a GlobalValueKey to compare by equality (using operator ==).

For example:

Key keyA = GlobalValueKey('1' + '23');
Key keyB = GlobalValueKey('12' + '3');

keyA == keyA; // true
keyA == keyB; // also true

If your key value is a String, you can also use a GlobalStringKey:

Key keyA = GlobalStringKey('1' + '23');
Key keyB = GlobalStringKey('12' + '3');

keyA == keyA; // true
keyA == keyB; // also true

Advanced #

There are many use cases for the GlobalValueKey, but I'd like to point out two in particular:

1) Creating keys inside the build method #

Flutter's documentation for global keys states that:

/// Creating a new GlobalKey on every build will throw away the state of the
/// subtree associated with the old key and create a new fresh subtree for the
/// new key. Besides harming performance, this can also cause unexpected
/// behavior in widgets in the subtree. For example, a [GestureDetector] in the
/// subtree will be unable to track ongoing gestures since it will be recreated
/// on each build.
///
/// Instead, a good practice is to let a State object own the GlobalKey, and
/// instantiate it outside the build method, such as in [State.initState].

However, this is only correct for Flutter's native GlobalObjectKey. The keys provided here in this package can indeed be recreated on every build with no problems (as long, of course, as the value you use to create the key has a well-behaved operator ==).

2) Global keys from data classes #

Suppose you want to derive keys from data classes, and then find widgets that correspond to them. For example, you have a User data class:

class User {
  final String name;
  User(this.name);  
  bool operator ==(Object other) => identical(this, other) || other is User && runtimeType == other.runtimeType && name == other.name;
  int get hashCode => name.hashCode;
}

Then you create a widget class from it:

class UserWidget extends StatefulWidget {
  final User user;
  UserWidget({required this.user});
  State<UserWidget> createState() => UserWidgetState();
}

class UserWidgetState extends State<UserWidget> {
  Widget build(BuildContext context) => Text(widget.user.name);  
}

To be able to find the user widget in the tree, modify your constructor as to create keys automatically, from the user:

UserWidget({required this.user}) : super(key: GlobalValueKey<UserWidgetState>(user));

And then, create a static method for easy access to the widget state:

static UserWidgetState? currentState(User user) =>
  GlobalValueKey<UserWidgetState>(user).currentState;

You now have easy access to your widget state, from anywhere: UserWidget.currentState(someUser)

You can find a complete working example here.

Some thoughts about global keys #

Global keys much more powerful than local keys, but in general, should not be used extensively. They are workarounds. I guess that's why Flutter does not provide a GlobalValueKey out of the box: Not to make global keys even more useful. Flutter has this philosophy that things that are useful, but can be used wrong, should be made difficult. In any case, I found that GlobalValueKey can sometimes make complex code orders of magnitude simpler.

MaskFunctionTextInputFormatter #

The MaskFunctionTextInputFormatter is a special TextInputFormatter that lets you format the text in a TextField or TextFormField as the user types, according to a mask; as well as also change that mask according to what is typed.

To use it, create your formatter and pass it to a TextField or TextFormField:

TextField(
  inputFormatters: [myFormatter],
);

This code is adapted from another package called mask_text_input_formatter by Sergey. The difference here is that instead of providing a mask, you provide a maskFunction that can change the mask automatically as the user types.

The maskFunction is of type MaskFunction:

typedef MaskFunction = String? Function({
   required TextEditingValue oldValue,
   required TextEditingValue newValue,
});

For example, suppose you want to format the text as ###### (where each # is a number) while the user typed less than 7 numbers, but you want to format it as ###.###.###-## for more chars than that:

var myFormatter = MaskFunctionTextInputFormatter(mask: _myFormatter);

String? _myFormatter({
   required TextEditingValue oldValue,
   required TextEditingValue newValue,
}) {
   if (newValue.text.length <= 6) return '######';
   else return '###.###.###-##';
}

When you create the formatter, you can also provide a filter parameter, to define the possible characters in the mask. If you don't provide the filter parameter, the default is that # matches a number, and A matches a letter:

var myFormatter = MaskFunctionTextInputFormatter(
   mask: _myFormatter,
   filter: {"#": RegExp('[0-9]'), "A": RegExp('[^0-9]')});

The getMaskedText and getUnmaskedText methods can be used if necessary:

// Get masked text:
print(maskFormatter.getMaskedText()); // -> "+0 (123) 456-78-90"

// Get unmasked text:
print(maskFormatter.getUnmaskedText()); // -> 01234567890

Important: Once again, please note all the above code is based upon mask_text_input_formatter by Sergey. The ONLY thing I added was the possibility of using a function that changes the mask. All the rest is from the original package, and credit belongs to their authors.

SideBySide #

The SideBySide widget disposes 2 widgets horizontally, while achieving a layout which is impossible for both the native Row and the RowSuper widgets.

The startChild will be on the left, and will occupy as much space as it wants, up to the available horizontal space.
The endChild will be on the right of the startChild widget, and it will occupy the rest of the available space. Note: This means, if the startChild widget occupies all the available space, then endChild widget will not be displayed (since it will be sized as 0 width).

For example, suppose you want to create a title with a divider that occupies the rest of the space:

return SideBySide(
  startChild: Text("First Chapter", textWidthBasis: TextWidthBasis.longestLine),
  endChild: Divider(color: Colors.grey),
  innerDistance: 8,
  minEndChildWidth: 20,
);

You can add an innerDistance, in pixels, between the startChild and endChild. The default is zero. It can be negative, in which case the widgets will overlap. The innerDistance is only used if the endChild is actually displayed.

You can define the minEndChildWidth, which is the minimum width, in pixels, that the endChild should occupy. The default is zero.

The crossAxisAlignment parameter specifies how to align the startChild and endChild vertically. The default is to center them. At the moment, only CrossAxisAlignment.start, CrossAxisAlignment.end and CrossAxisAlignment.center work.

Try running the SideBySide example.

Button #

The Button widget transforms any widget into a button with some immediate visual response to a tap. It provides you with a builder, and the isPressed boolean which tells you if the widget is being touched or not. It can also expand the click-area, show the click-area (for debug purposes), sustain the visual effect of the tap for some duration, and accepts a throttle period between taps:

Button(
    onTap: () {...},
    minVisualTapDuration: Duration(milliseconds: 200),
    tapThrottle: Duration(milliseconds: 500),
    clickAreaMargin: const Pad(horizontal: 40.0, vertical: 20),
    debugShowClickableArea: true,
    builder: ({required bool isPressed}) => 
      Text('Click Me', style: TextStyle(color: isPressed ? Colors.black : Colors.white)),        
);

When the user taps the button, the isPressed boolean will be true for at least minVisualTapDuration.
The widget will only feel another tap if tapThrottle duration has passed since the last tap.
The click-area can be expanded by a margin given by clickAreaMargin, thus making the widget easier to tap.
If debugShowClickableArea is true, the click-area will be shown in red.

Try running the Button and CircleButton example.

CircleButton #

The CircleButton is similar to Flutter's native IconButton, but with a few differences. Since circular buttons are small, the user's finger usually hides it during the tap. For this reason, the CircleButton will:

Show an immediate visual feedback to a tap, and then sustain that feedback for about 100 milliseconds, enough time for the user to remove the finger and see it.
You can expand the click-area, to make the button easier to tap. You can also show the click-area, for debug purposes.

CircleButton(
    onTap: () {...},
    icon: Icon(Icons.shopping_cart, color: Colors.white),
    clickAreaMargin: const Pad(left: 30, right: 50, vertical: 20),
    debugShowClickableArea: true,
    backgroundColor: Colors.white30,
    tapColor: Colors.black,
    border: Border.all(width: 1, color: Colors.black),
    size: 56,                
);

When the user taps the button, the isPressed boolean will be true for at least minVisualTapDuration.
The widget will only feel another tap if tapThrottle duration has passed since the last tap.
The click-area can be expanded by a margin given by clickAreaMargin, thus making the widget easier to tap.
If debugShowClickableArea is true, the click-area will be shown in red.

Try running the Button and CircleButton example.

CaptureGestures #

A widget that captures gestures, preventing its parent (and ascending subtree) to feel them.

How is this different from IgnorePointer and AbsorbPointer?

IgnorePointer makes itself and its child (and the descending subtree) invisible to touches. This means, for example, if you put the IgnorePointer and its child above some widget inside a Stack, the touches will "pass through" the IgnorePointer and be felt by the widget below it. The gesture can also be felt by the IgnorePointer's parent.

AbsorbPointer also makes its child invisible to touches, but it will also prevent the touch to be felt by widgets below it in a Stack. But the gesture can be felt by the AbsorbPointer's parent.

As you can see, IgnorePointer and AbsorbPointer act on their children and on widgets below them in a Stack. However, in both cases the gesture can be felt by their parents.

For example, if you put IgnorePointer or AbsorbPointer inside a ListView, none of them will prevent the ListView to be scrolled, because the ListView is in the ascending subtree).

The CaptureGestures however, will let its child feel the touches it cares about, and then capture and cancel other touches that reach the CaptureGestures itself. This means the CaptureGestures parent (and all the ascending subtree) will not feel the touches below the CaptureGestures area.

The parameters for the CaptureGestures.only() constructor are:

capturingTap turns on/off the capturing of tap gestures.
capturingDoubleTap turns on/off the capturing of double-tap gestures.
capturingLongPress turns on/off the capturing of long-press gestures.
capturingVerticalDrag turns on/off the capturing of vertical-drag gestures.
capturingHorizontal turns on/off the capturing of horizontal-drag gestures.
capturingForcePress turns on/off the capturing of force-press gestures.
display prints the captured events to the console (for debug reasons only).

The CaptureGestures.all() constructor will capture all gestures.

The CaptureGestures.tap() constructor will capture only single tap gestures.

Preventing scroll #

The CaptureGestures.scroll() constructor will capture only drag gestures (preventing both vertical and horizontal scroll).

Consider this code:

ListView(
   children: [
      CaptureGestures.scroll(child: ElevatedButton(...)),
      ...   
      ...   
   ]
);

Here, the ElevatedButton can feel the onTap gesture that it cares about. But the user cannot scroll the list by touching the button and dragging it up or down, because the drag gestures are getting captured by the CaptureGestures, never reaching the ListView.

Note: Setting the scrollable's physics to NeverScrollableScrollPhysics() is also an option, but sometimes you can't do that. Also, CaptureGestures allows you to choose just a part of the widget tree to cancel the scroll.

Try running the CaptureGestures example.

NonUniformOutlineInputBorder #

Can be used to style text-fields and containers.

Similar to Flutter's native OutlineInputBorder but you can hide some of the sides, by setting hideTopSide, hideBottomSide, hideRightSide and hideLeftSide to true.

Usage for text-fields:

TextField(
   decoration: InputDecoration(
      enabledBorder: NonUniformOutlineInputBorder(hideLeftSide: true, ...),
      ...

Usage for containers:

Container(
   decoration: ShapeDecoration(
      shape: NonUniformOutlineInputBorder(
         hideLeftSide: true,
         borderSide: BorderSide(width: 10),
         borderRadius: BorderRadius.only(
            topLeft: Radius.circular(30),
            topRight: Radius.circular(15),
            bottomLeft: Radius.circular(10),
            bottomRight: Radius.circular(35),
         ),          
   ...

Try running the NonUniformOutlineInputBorder example.

NonUniformRoundedRectangleBorder #

This may be used to style buttons and containers.

Similar to Flutter's native RoundedRectangleBorder but you can hide some of the sides, by setting hideTopSide, hideBottomSide, hideRightSide and hideLeftSide to false.

Usage for buttons:

ElevatedButton(
   style: ElevatedButton.styleFrom(
      shape: NonUniformRoundedRectangleBorder(
         hideLeftSide: true,
         side: BorderSide(color: Colors.black87, width: 15.0),
         borderRadius: BorderRadius.only(
            topLeft: Radius.circular(30),
            topRight: Radius.circular(15),
            bottomLeft: Radius.circular(10),
            bottomRight: Radius.circular(35),
         ), 
      ...