Select2 4.0.0

The 4.0 release of Select2 is the result of three years of working on the code base and watching where it needs to go. At the core, it is a full rewrite that addresses many of the extensibility and usability problems that could not be addressed in previous versions.

This release contains many breaking changes, but easy-upgrade paths have been created as well as helper modules that will allow for backwards compatibility to be maintained with past versions of Select2. Upgrading will require you to read the release notes carefully, but the migration path should be relatively straightforward. You can view a list of the most common changes that you will need to make in the release notes.

Below is an in-depth review of what is new in Select2, as well as some of the major changes that have been made.

New features

The notable features of this new release include:

  • A more flexible plugin framework that allows you to override Select2 to behave exactly how you want it to.
  • Consistency with standard <select> elements for all data adapters, removing the need for hidden <input> elements.
  • A new build system that uses AMD to keep everything organized.
  • Less specific selectors allowing for Select2 to be styled to fit the rest of your application.

Plugin system

Select2 now provides interfaces that allow for it to be easily extended, allowing for anyone to create a plugin that changes the way Select2 works. This is the result of Select2 being broken into four distinct sections, each of which can be extended and used together to create your unique Select2.

The adapters implement a consistent interface that is documented in the options section for adapters, allowing you to customize Select2 to do exactly what you are looking for. Select2 is designed such that you can mix and match plugins, with most of the core options being built as decorators that wrap the standard adapters.

AMD-based build system

Select2 now uses an AMD-based build system, allowing for builds that only require the parts of Select2 that you need. While a custom build system has not yet been created, Select2 is open source and will gladly accept a pull request for one.

Select2 includes the minimal almond AMD loader, but a custom select2.amd.js build is available if you already use an AMD loader. The code base (available in the src directory) also uses AMD, allowing you to include Select2 in your own build system and generate your own builds alongside your existing infrastructure.

The AMD methods used by Select2 are available as jQuery.fn.select2.amd.define()/require(), allowing you to use the included almond loader. These methods are primarily used by the translations, but they are the recommended way to access custom modules that Select2 provides.

Migrating from Select2 3.5

There are a few breaking changes that migrators should be aware of when they are coming from older versions of Select2.

If you use the full build of Select2 (select2.full.js), you will be automatically notified of the major breaking changes, and compatibility modules will be used in some cases to ensure that your code still behaves how you were expecting.

No more hidden input tags

In past versions of Select2, an <input type="hidden" /> tag was recommended if you wanted to do anything advanced with Select2, such as work with remote data sources or allow users to add their own tags. This had the unfortunate side-effect of servers not receiving the data from Select2 as an array, like a standard <select> element does, but instead sending a string containing the comma-separated strings. The code base ended up being littered with special cases for the hidden input, and libraries using Select2 had to work around the differences it caused.

In Select2 4.0, the <select> element supports all core options, and support for the old <input type="hidden" /> has been deprecated. This means that if you previously declared an AJAX field with some pre-selected options that looked like…

1
<input type="hidden" name="select-boxes" value="1,2,4,6" />

It will need to be recreated as a <select> element with some <option> tags that have value attributes that match the old value.

1
2
3
4
5
6
<select name="select-boxes" multiple="multiple">
  <option value="1" selected="selected">Select2</option>
  <option value="2" selected="selected">Chosen</option>
  <option value="4" selected="selected">selectize.js</option>
  <option value="6" selected="selected">typeahead.js</option>
</select>

The options that you create should have selected="selected" set so Select2 and the browser knows that they should be selected. The value attribute of the option should also be set to the value that will be returned from the server for the result, so Select2 can highlight it as selected in the dropdown. The text within the option should also reflect the value that should be displayed by default for the option.

Advanced matching of searches

In past versions of Select2, when matching search terms to individual options, which limited the control that you had when displaying results, especially in cases where there was nested data. The matcher function was only given the individual option, even if it was a nested options, without any context.

With the new matcher function, only the root-level options are matched and matchers are expected to limit the results of any children options that they contain. This allows developers to customize how options within groups can be displayed, and modify how the results are returned.

A function has been created that allows old-style matcher functions to be converted to the new style. You can retrieve the function from the select2/compat/matcher module, which should just wrap the old matcher function.

So if your old code used a matcher that only displayed options if they started with the term that was entered, it would look something like…

1
2
3
4
5
6
7
8
9
10
11
function matchStart (term, text) {
  if (text.toUpperCase().indexOf(term.toUpperCase()) == 0) {
    return true;
  }

  return false;
}

$("select").select2({
  matcher: matchStart
})

Then in Select2 4.0, you would need to wrap the matchStart method (or the name of the matcher you created) with a oldMatcher method that we have created.

1
2
3
4
5
6
7
8
9
10
11
12
13
function matchStart (term, text) {
  if (text.toUpperCase().indexOf(term.toUpperCase()) == 0) {
    return true;
  }

  return false;
}

$.fn.select2.amd.require(['select2/compat/matcher'], function (oldMatcher) {
  $("select").select2({
    matcher: oldMatcher(matchStart)
  })
});

This will work for any matchers that only took in the search term and the text of the option as parameters. If your matcher relied on the third parameter containing the jQuery element representing the original <option> tag, then you may need to slightly change your matcher to expect the full JavaScript data object being passed in instead. You can still retrieve the jQuery element from the data object using the data.element property.

More flexible placeholders

In the most recent versions of Select2, placeholders could only be applied to the first (typically the default) option in a <select> if it was blank. The placeholderOption option was added to Select2 to allow users using the select tag to select a different option, typically an automatically generated option with a different value.

The placeholder option can now take an object as well as just a string. This replaces the need for the old placeholderOption, as now the id of the object can be set to the value attribute of the <option> tag.

For a select that looks like the following, where the first option (with a value of -1) is the placeholder option…

1
2
3
4
<select>
  <option value="-1" selected="selected">Select an option</option>
  <option value="1">Something else</option>
</select>

You would have previously had to get the placeholder option through the placeholderOption, but now you can do it through the placeholder option by setting an id.

1
2
3
4
5
6
$("select").select2({
  placeholder: {
    id: "-1",
    placeholder: "Select an option"
  }
})

And Select2 will automatically display the placeholder when the value of the select is -1, which it will be by default. This does not break the old functionality of Select2 where the placeholder option was blank by default.

Display reflects the actual order of the values

In past versions of Select2, choices were displayed in the order that they were selected. In cases where Select2 was used on a <select> element, the order that the server received the selections did not always match the order that the choices were displayed, resulting in confusion in situations where the order is important.

Select2 will now order selected choices in the same order that will be sent to the server.

Changed method and option names

When designing the future option set for Select2 4.0, special care was taken to ensure that the most commonly used options were brought over. For the most part, the commonly used options of Select2 can still be referenced under their previous names, but there were some changes which have been noted.

Removed the requirement of initSelection

In the past, whenever you wanted to use a custom data adapter, such as AJAX or tagging, you needed to help Select2 out in determining the initial values that were selected. This was typically done through the initSelection option, which took the underlying data of the input and converted it into data objects that Select2 could use.

This is now handled by the data adapter in the current method, which allows Select2 to convert the currently selected values into data objects that can be displayed. The default implementation converts the text and value of option elements into data objects, and is probably suitable for most cases. An example of the old initSelection option is included below, which converts the value of the selected options into a data object with both the id and text matching the selected value.

1
2
3
4
5
6
7
8
9
{
  initSelection : function (element, callback) {
    var data = [];
    $(element.val()).each(function () {
      data.push({id: this, text: this});
    });
    callback(data);
  }
}

When using the new current method of the custom data adapter, this method is called any time Select2 needs a list of the currently selected options. This is different from the old initSelection in that it was only called once, so it could suffer from being relatively slow to process the data (such as from a remote data source).

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
$.fn.select2.amd.require([
  'select2/data/array',
  'select2/utils'
], function (ArrayData, Utils) {
  function CustomData ($element, options) {
    CustomData.__super__.constructor.call(this, $element, options);
  }

  Utils.Extend(CustomData, ArrayData);

  CustomData.prototype.current = function (callback) {
    var data = [];
    var currentVal = this.$element.val();

    if (!this.$element.prop('multiple')) {
      currentVal = [currentVal];
    }

    for (var v = 0; v < currentVal.length; v++) {
      data.push({
        id: currentVal[v],
        text: currentVal[v]
      });
    }

    callback(data);
  };

  $("#select").select2({
    dataAdapter: CustomData
  });
}

The new current method of the data adapter works in a similar way to the old initSelection method, with three notable differences. The first, and most important, is that it is called whenever the current selections are needed to ensure that Select2 is always displaying the most accurate and up to date data. No matter what type of element Select2 is attached to, whether it supports a single or multiple selections, the data passed to the callback must be an array, even if it contains one selection. The last is that there is only one parameter, the callback to be executed with the latest data, and the current element that Select2 is attached to is available on the class itself as this.$element.

If you only need to load in the initial options once, and otherwise will be letting Select2 handle the state of the selections, you don't need to use a custom data adapter. You can just create the <option> tags on your own, and Select2 will pick up the changes.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
var $element = $('select').select2(); // the select element you are working with

var $request = $.ajax({
  url: '/my/remote/source' // wherever your data is actually coming from
});

$request.then(function (data) {
  // This assumes that the data comes back as an array of data objects
  // The idea is that you are using the same callback as the old `initSelection`

  for (var d = 0; d < data.length; d++) {
    var item = data[d];

    // Create the DOM option that is pre-selected by default
    var option = new Option(item.text, item.id, true, true);

    // Append it to the select
    $element.append(option);
  }

  // Update the selected options that are displayed
  $element.trigger('change');
});

Custom data adapters instead of query

In the past, any time you wanted to hook Select2 up to a different data source you would be required to implement custom query and initSelection methods. This allowed Select2 to determine the initial selection and the list of results to display, and it would handle everything else internally, which was fine more most people.

The custom query and initSelection methods have been replaced by custom data adapters that handle how Select2 stores and retrieves the data that will be displayed to the user. An example of the old query option is provided below, which is the same as the old example, and it generates results that contain the search term repeated a certain number of times.

1
2
3
4
5
6
7
8
9
10
11
{
  query: function (query) {
    var data = {results: []}, i, j, s;
    for (i = 1; i < 5; i++) {
      s = "";
      for (j = 0; j < i; j++) {s = s + query.term;}
      data.results.push({id: query.term + i, text: s});
    }
    query.callback(data);
  }
}

This has been replaced by custom data adapters which define a similarly named query method. The comparable data adapter is provided below as an example.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
$.fn.select2.amd.require([
  'select2/data/array',
  'select2/utils'
], function (ArrayData, Utils) {
  function CustomData ($element, options) {
    CustomData.__super__.constructor.call(this, $element, options);
  }

  Utils.Extend(CustomData, ArrayData);

  CustomData.prototype.query = function (params, callback) {
    var data = {
      results: []
    };

    for (var i = 1; i < 5; i++) {
      var s = "";

      for (var j = 0; j < i; j++) {
        s = s + params.term;
      }

      data.results.push({
        id: params.term + i,
        text: s
      });
    }

    callback(data);
  };

  $("#select").select2({
    dataAdapter: CustomData
  });
}

The new query method of the data adapter is very similar to the old query option that was passed into Select2 when initializing it. The old query argument is mostly the same as the new params that are passed in to query on, and the callback that should be used to return the results is now passed in as the second parameter.

Renamed templating options

Select2 previously provided multiple options for formatting the results list and selected options, commonly referred to as "formatters", using the formatSelection and formatResult options. As the "formatters" were also used for things such as localization, which has also changed, they have been renamed to templateSelection and templateResult and their signatures have changed as well.

You should refer to the updated documentation on templates when migrating from previous versions of Select2.

The id and text properties are strictly enforced

When working with array and AJAX data in the past, Select2 allowed a custom id function or attribute to be set in various places, ranging from the initialization of Select2 to when the remote data was being returned. This allowed Select2 to better integrate with existing data sources that did not necessarily use the id attribute to indicate the unique identifier for an object.

Select2 no longer supports a custom id or text to be used, but provides integration points for converting incorrect data to the expected format.

When working with array data

Select2 previously supported defining array data as an object that matched the signature of an AJAX response. A text property could be specified that would map the given property to the text property on the individual objects. You can now do this when initializing Select2 by using the following jQuery code to map the old text and id properties to the new ones.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
var data = $.map([
  {
    pk: 1,
    word: 'one'
  },
  {
    pk: 2,
    word: 'two'
  }
], function (obj) {
  obj.id = obj.id || obj.pk;
  obj.text = obj.text || obj.word;

  return obj;
});

This will result in an array of data objects that have the id properties that match the existing pk properties and text properties that match the existing word properties.

When working with remote data

The same code that was given above can be used in the processResults method of an AJAX call to map properties there as well.

Renamed translation options

In previous versions of Select2, the default messages provided to users could be localized to fit the language of the website that it was being used on. Select2 only comes with the English language by default, but provides community-contributed translations for many common languages. Many of the formatters have been moved to the language option and the signatures of the formatters have been changed to handle future additions.

Declaring options using data-* attributes

In the past, Select2 has only supported declaring a subset of options using data-* attributes. Select2 now supports declaring all options using the attributes, using the format specified in the documentation.

You could previously declare the URL that was used for AJAX requests using the data-ajax-url attribute. While Select2 still allows for this, the new attribute that should be used is the data-ajax--url attribute. Support for the old attribute will be removed in Select2 4.1.

Although it was not documented, a list of possible tags could also be provided using the data-select2-tags attribute and passing in a JSON-formatted array of objects for tags. As the method for specifying tags has changed in 4.0, you should now provide the array of objects using the data-data attribute, which maps to the array data option. You should also enable tags by setting data-tags="true" on the object, to maintain the ability for users to create their own options as well.

If you previously declared the list of tags as…

1
<select data-select2-tags='[{"id": "1", "text": "One"}, {"id": "2", "text": "Two"}]'></select>

…then you should now declare it as…

1
<select data-data='[{"id": "1", "text": "One"}, {"id": "2", "text": "Two"}]' data-tags="true"></select>

Deprecated and removed methods

As Select2 now uses a <select> element for all data sources, a few methods that were available by calling .select2() are no longer required.

.select2("val")

The "val" method has been deprecated and will be removed in Select2 4.1. The deprecated method no longer includes the triggerChange parameter.

You should directly call .val on the underlying <select> element instead. If you needed the second parameter (triggerChange), you should also call .trigger("change") on the element.

1
$("select").val("1").trigger("change"); // instead of $("select").select2("val", "1");

.select2("enable")

Select2 will respect the disabled property of the underlying select element. In order to enable or disable Select2, you should call .prop('disabled', true/false) on the <select> element. Support for the old methods will be completely removed in Select2 4.1.

1
$("select").prop("disabled", true); // instead of $("select").enable(false);