ES6 and beyond part 1

written by Maurits Lamers

Introduction

This is the first part of a series of blog posts in which I want to show my attempts in trying to see in what way SproutCore can be rewritten in more modern JavaScript without changing the essential concepts or syntax too much.

This also provides an opportunity to show a few concepts that make SproutCore what it is and show how they are implemented.

Since the beginning of SproutCore in 2007-2008 JavaScript has developed hugely and changed dramatically. The most eye catching changes came with ES6, such as classes, arrow functions, and the module system. At the same time also the way developers dealt with the runtime environment changed. Many of the early frameworks dealt with problems by changing the prototypes of standard JS objects, such as Object, Array etc. Nowadays this behavior is frowned upon, mainly because of the conflicts this can cause when multiple frameworks meet in a page.

From the start SproutCore was conceived as a Cocoa like framework, in the sense that the type of API it provides very much resembles that of a desktop framework. It provides a series of Views, a Model and Controller system and to make it all work smoothly a Key Value Observing (KVO) system.

In this post I want to mostly concentrate on the KVO system and the way this has been implemented in a classical inheritance scheme.

SproutCore still alive and kicking

written by admin

It has been quite a while since the last blog post. The reason for it is that there have been quite a few changes taking place since the last post.

The first news is that Tyler Keating has stepped down as project lead and has appointed Maurits Lamers to take over. From this spot we want to thank Tyler for the enormous amount of work he has done for the project and we wish him good luck.

Secondly we have been busy building new build tools using NodeJS instead of Ruby. The new build tools are simply called BT and are written in SproutCore. You can find more info on BT in the repository: https://github.com/sproutcore/build-tools. As the BT has had quite some time to mature, we are now dropping official support for the old Ruby based Abbot. The new BT offers quite a few advantages over Abbot:

  • Development server is much faster: The BT keeps and transforms all files in memory, speeding up a reload from 10s down to around 2s.

  • Many different build types: In abbot, it always was difficult to build SproutCore apps for different environments, such as Cordova. BT adds support for different types of bundling to match your specific wishes.

  • ES5 strict compatible: Abbot only supported SproutCore builds in ES5 non-strict mode. BT allows your SproutCore app to be fully ES5 strict compatible, and also offers support for ES6 through Traceur.

  • Written in SproutCore: BT is written in SproutCore, so as a developer familiar to SproutCore, it is much easier to contribute.

Thirdly the community has lost a few active core team members who were hired away. We want to thank them for their contributions. Sadly, the downside is that it has left a few projects such as documentation updates that were underway without active support. As such we are very much welcoming contributers of any kind!

Thank you for your patience and continuing support!

SproutCore 1.11.0 Release

written by admin

The SproutCore team is very pleased to announce the final release of SproutCore version 1.11.0. This version continues in the tradition of SproutCore releases to become the fastest, most feature-rich and most stable version of SproutCore to-date and includes several new additions and improvements to make development with SproutCore even easier and to make SproutCore apps even more impressive.

For those wishing to install SproutCore for the first time, please visit http://sproutcore.com/install/ for instructions. If you are upgrading from a previous version of SproutCore, simply run the following:

<code>$ gem update sproutcore
</code>

If you have any trouble installing or upgrading to 1.11, be sure to visit the mailing list sproutcore@googlegroups.com or the IRC channel #sproutcore for support. As always, the team and community are here to help!

Highlights of Version 1.11.0

The majority of the work done in 1.11.0 was to improve the existing API and to fix bugs. The list of changes is too long to include, but the following are the highlights.

Major New Features

Polymorphic Records (SC.Record.prototype.isPolymorphic)

The experimental polymorphic record support has been rewritten. It is now bug free, significantly faster and included in the datastore framework by default. This allows you to support truly polymorphic record types in the store. For more information on using polymorphism, there is a detailed description and useful examples in the SC.Record documentation here: http://docs.sproutcore.com/#doc=SC.Record&src=false.

SC.Binding Transforms

There are a few new binding transforms that can be used to reduce the amount of code you as a developer need to write. These are string, integer, equalTo and mix. The transforms string and integer are very simple and simply transform the value into a String or an Integer (according to radix argument). This is useful in particular for custom views that don’t want have to coerce every possible defined property value into a String or an Integer when necessary. See here: http://docs.sproutcore.com/#doc=SC.Binding&method=string&src=false and here: http://docs.sproutcore.com/#doc=SC.Binding&method=integer&src=false.

The equalTo transform is also very simple and simply returns true or false depending on whether the bound value is equal to the argument value. See here: http://docs.sproutcore.com/#doc=SC.Binding&method=equalTo&src=false.

Finally, the mix transform is a more powerful transform that may take some research to use properly. The benefit of mix is that it allows you to mix together multiple bound values into a single transform function. This is similar to the way that the and and or transforms work (in fact these were refactored to use mix), but mix allows you to specify a custom transform function. See here: http://docs.sproutcore.com/#doc=SC.Binding&method=mix&src=false.

Cross-origin resource sharing (CORS) with credentials

To allow SC.Request to work properly with cross-site Access-Control requests, we’ve added a new property to SC.Request, allowCredentials. This will allow cookies and authorization headers to be sent by configuring the withCredentials property of the XHR request. As part of this change, SC.Request includes a new property, isSameDomain, which correctly identifies if a request is cross-origin or not and only attempts to set the withCredentials property when not targeting the same domain. A brief description of this property is available here: http://docs.sproutcore.com/#doc=SC.Request&method=allowCredentials&src=false.

SC.WebSocket

To simplify using WebSockets and to ensure that WebSocket messaging triggers a run of the SproutCore run loop, we’ve added a new class SC.WebSocket. Using SC.WebSocket makes communicating with WebSockets trivial to add to any SproutCore app. For details and examples, please see the documentation here: http://docs.sproutcore.com/#doc=SC.WebSocket&src=false.

SC.NestedStore

The nested store is an extremely powerful feature of SproutCore that allows for complex data management within the client. For 1.11, nested stores have been expanded slightly to support even more advanced uses. The first addition is a new property conflictedStoreKeys that contains a list of all store keys that are in conflict with the main store. The purpose of this is for applications that automatically update shared live data that multiple users may edit. In such a scenario, one user might be editing a record that another user already completed and saved. Using conflictedStoreKeys the application can check for the presence of any conflicts from new data coming in before attempting to commit the changes from the nested store. Here: http://docs.sproutcore.com/#doc=SC.NestedStore&method=conflictedStoreKeys&src=false.

The other addition is the new SC.Store.prototype.chainAutonomousStore() method, which returns a nested store that is allowed to interact directly with a remote data store. Typically, nested stores are prohibited from committing changes directly to a remote data store, instead having to commit them to the main store which would then forward the changes on to the remote data store (via a data source which handles the actual communication and integration between client and server). However, by using an autonomous nested store, you can choose to commit changes within the autonomous nested store and only push successful changes to the main store. This can simplify the process when commits to the remote are expected to fail occasionally and we wish to avoid pushing any possibly invalid data into the main store. Here: http://docs.sproutcore.com/#doc=SC.Store&method=chainAutonomousStore&src=false.

Scale and Transform Origin

Scale is now a first-class layout property, correctly impacting frame and clippingFrame. If a view is scaled, the width & height of the frame will be correct as the view appears. For example, a view with layout equal to { width: 100, height: 100, scale: 2 } will report a frame of { x: 0, y: 0, width: 200, height: 200, scale: 2 }. The clippingFrame also takes a scaling origin into account as well and as part of this change, there are two new layout properties: transformOriginX and transformOriginY, which define the percentage (between 0.0 and 1.0) on the respective axis about which the scale transform is applied. These properties affect all transform styles and so can be used to also change the origin of a rotate style. You can learn about using layouts here: http://docs.sproutcore.com/#doc=SC.View&method=layout&src=false.

Legacy Framework

SproutCore will work with browsers as old as IE7 (don’t laugh, we all have customers that can not/will not move to newer browsers for some time), however in order to prepare for the day when the lowest common denominator can be raised, there is a new sub-framework within SproutCore, :'sproutcore/legacy', which is meant to contain all code providing support for legacy browsers. Currently, this includes the existing polyfill for window.requestAnimationFrame and a new polyfill for Object.keys, but will continue to grow to include any specific workarounds for these older browsers. Because it’s important for SproutCore apps to “just work” as much as possible, the legacy framework is included by default by requiring :sproutcore in a project’s or an app’s Buildfile.

To build an app that will only work with the newest browsers (probably not a great idea —), you may change your Buildfile requirements to include only the specific SproutCore sub-frameworks you need. For example,

<code>config :my_app, :required => [:"sproutcore/desktop", :"sproutcore/datastore"]
</code>

Major Changes & Improvements

SC.Gesturable & SC.Gesture (SC.TapGesture, SC.PinchGesture, SC.SwipeGesture) Refactor

After some investigation, it was found that there were a number of issues with the built-in gesture support. For example, two touch taps would throw an exception, pinches would fail to register and in particular, supporting multiple gestures concurrently failed in a number of scenarios. In order to fix these problems, the gesture code has been rewritten from the top-down.

It is now possible to mixin SC.Gesturable to a view and use events to react to multiple gestures concurrently. Examples of the advanced type of behavior that the gesture code allows include,

  • Responding to single finger, two finger or any other number of touch taps, pinches (> 2 touches) or swipes individually or as a group. For example, your code may want to perform different actions when a single finger taps vs. when there is a two finger tap.

  • A touch session, the time between when the first touch begins and the last touch ends, may contain more than one gesture. For example, it’s possible for the user to perform a pinch, then use a third finger to tap, then swipe the remaining fingers. At the least, the ability to perform gestures in a single touch session multiple times, makes the gesture recognition more robust against stray accidental touches.

  • Swipe gestures can now be configured to match against any combination of arbitrary angles, not just combinations of left, right, up & down.

  • Swipe gestures no longer trigger by simply moving far enough in one direction. They must also move quickly and end quickly (configurable).

How does this affect your code?

For the most part, this should have no effect on existing implementors of SC.Gesturable. The three built-in gestures: SC.TapGesture, SC.PinchGesture, and SC.SwipeGesture are still defined and they work much better than before. However if you have defined a custom SC.Gesture subclass, it will unfortunately not work correctly with this update. Because we felt the previous version of SC.Gesture’s API was too complex and incompatible with the behavior we needed to achieve, we decided it was better to rewrite it in a simpler form. We’re very sorry for this backwards incompatibility, and would be happy to help with a migration (see mailing list and IRC info above).

SC.PickerPane Pop “Out of the Anchor”

This view has been given special behavior when used with SC.View’s transitionIn plugin support. If the plugin defines layoutProperties of either scale or rotate, then the picker will adjust its transform origin X & Y position to appear to scale or rotate out of the anchor. The result is a very nice effect that picker panes appear to pop right out of their anchors, rather than just appearing offset. To see it in effect, simply set the transitionIn property of the pane to one of SC.View.SCALE_IN or SC.View.POP_IN.

SC.ScrollView Refactor

The most ambitious undertaking in 1.11, was a total refactor of SC.ScrollView. There were a few features that we wanted to fix and improve in scroll views, such as scaling, alignment and touch handling and after a few attempts, we rebuilt it entirely from scratch method-by-method. This allowed us to question each design decision that went into the original SC.ScrollView and search for improvements and the result is that SC.ScrollView works better and is faster then ever before. We were able to reduce the amount of code, remove some internal observers and still do a better job of supporting touch, scale and alignment.

For example, scaling the content of a scroll view will now maintain its visual center rather than hugging the side, which dramatically improves the zooming experience. Touch gesture animations, such as the bounce back and deceleration, are now animated using requestAnimationFrame for smoother performance.

Here is a list of important changes,

  • SC.ScrollView.prototype.scale now works as advertised!

  • If horizontalOverlay or verticalOverlay is true, the scroll view will use SC.OverlayScrollerView by default. It’s no longer necessary to also set horizontalScrollerView: SC.OverlayScrollerView in order to get overlaid scrollers.

  • Overlaid scroller bars now fade out when not in use.

  • Use horizontalAlign and verticalAlign properties to align fixed-width or -height content.

  • SC.ScrollView alignment handling has been improved for container (i.e. the scroll view) or content size changes to support the following scenarios:

  1. The scroll view is left (or top) aligned, scrolled to the maximum right (or bottom) edge and container or content changes size: the content should stick to the right (or bottom) side.

  2. The scroll view is left (or top) aligned, scrolled to the minimum left (or top) edge and container or content changes size: the content should stick to the left (or top) side.

  3. The scroll view is right (or bottom) aligned, scrolled to the maximum right (or bottom) edge and container or content changes size: the content should stick to the right (or bottom) side.

  4. The scroll view is right (or bottom) aligned, scrolled to the minimum left (or top) edge and container or content changes size: the content should stick to the left (or top) side.

  5. The scroll view is center (or middle) aligned, scrolled to the center (or middle) and container or content changes size: the content should stick to the center (or middle).

Finally, a critical difference in the new SC.ScrollView is that it uses SC.View layout changes to position its content rather than adding margins that pushed the content beyond what its layout was set to. The reason for this change was to allow for GPU accelerated content positioning without duplicating code (the SC.View layout code already supports it). This means that it is possible to fix the layout of your content (layout with top, left, width & height) and set wantsAcceleratedLayer to true to get your content scrolled via GPU accelerated positioning. This can improve each scroll event render by as much as 50%.

SC.MenuScrollView Refactor

As with the SC.ScrollView refactor, SC.MenuScrollView has been improved greatly in its implementation and performance.

Responsive Design and Design Modes

Design modes are the answer to the problem of how to create a SproutCore web app that can handle vastly different screen size “designs”, beyond just stretching and squishing, as simply as possible and without hurting performance. While CSS media queries may be used for static websites, they don’t provide much value to a client side app like those made with SproutCore, primarily because using CSS to change the layout and the visibility of views would break the state of the application that exists in code.

Instead, we use the modeAdjust property and add simple overrides to the layout or any other properties of views that will apply for the specific design “mode”. For instance, a view can be visible by default and only hidden in the small, ‘s’ mode with a simple hash,

<code>largeOnlyView: SC.View.extend({
  isVisible: true, // visible by default
  modeAdjust: { 's': { isVisible: false } } // hidden on small screens
})
</code>

For a demonstration of this, please visit the Showcase here: http://showcase.sproutcore.com/#demos/Responsive%20Design and for more details on using design modes, see here: http://docs.sproutcore.com/#doc=SC.Application&src=false.

General Changes & Fixes

There are are literally hundreds of other changes and fixes, which would take many more pages to write about. To see the entire list, please view the Change Log (Note that there were three release candidates with their respective changes listed below them). Here is a small sample of the many changes,

  • Improved: When the online/offline value of window changes, SC.device runs the run loop.

  • Improved: Subclasses can now override bindings without issue.

  • Added: DateTime localizations are now available in French.

  • Improved: Updated to jQuery 1.11.1. Notably, there was a serious memory leak in jQuery 1.8 where the tokenCache grew continuously.

  • Improved: SC.State performance by removing expensive observers.

  • Improved: Performance of core SC methods: SC.mixin, SC.supplement as well as SC.Function.enhance. Benchmark: SC.mixin & SC.supplement ~ 58% faster

  • Improved: Removed several instances of arguments instantiation, which is terribly slow.

  • Added: Added toolTip property to SC.MenuItemView, and the corresponding itemToolTipKey to SC.MenuView.

  • Added: Added horizontal layout for SC.ListView using layoutDirection property.

  • Improved: SC.UndoManager has been refactored to allow easier undo action grouping.

  • Added: SC.DateTime.prototype.toFormattedString now takes ‘%o’ formatter to specify the date’s ordinal.

And the list goes on…

SproutCore 1.11.0 Release Candidate 3

written by admin

The final release candidate for version 1.11.0 is now available.

This release candidate fixes a few of the remaining bugs and addresses the regressions that had crept in since RC1. It also introduces several improvements and nice new features, such as the ability for SC.PickerPanes to pop “out” of their anchors when using a scale transitionIn plugin (i.e. SC.View.SCALE_IN) for a really nice effect or the ability to easily check for conflicts between a nested store and its parent using the new conflictedStoreKeys property. For a list of all the changes, please view the detailed change log which can be seen here: SproutCore CHANGELOG.

To install SproutCore 1.11.0.rc3 for testing, please upgrade your previous version of SproutCore by running the following:

<code>gem update sproutcore --prerelease
</code>

We will be using your feedback over the next couple weeks to finalize 1.11.0, so please be sure to try it out and let us know what issues remain: Github Issues. To discuss the next version or to discuss SproutCore in general, as always please use #sproutcore on IRC or email to sproutcore@googlegroups.com.

SproutCore 1.11.0 Release Candidate 2

written by admin

The next release candidate for version 1.11.0 is now available.

This release candidate addresses a few bugs, makes some important under-the-hood improvements (e.g. SC.View layout update performance, reducing SC.Event memory churn, avoiding leaking arguments) and updates the API of SC.ActionSupport. The detailed change log can be viewed here: https://github.com/sproutcore/sproutcore/blob/master/CHANGELOG.md

To install SproutCore 1.11.0.rc2 for testing, please upgrade your previous version of SproutCore by running the following:

<code>gem update sproutcore --prerelease
</code>

We will be using your feedback over the next few weeks to finalize 1.11.0, so please be sure to try it out and let us know what issues remain: Github Issues. To discuss the next version or to discuss SproutCore in general, as always please use #sproutcore on IRC or email to sproutcore@googlegroups.com.