Platforms State of the Union 

Session 102 WWDC 2015

WWDC 2015 Platforms State of the Union

Ladies and gentlemen, please welcome Vice President OS X Platform, Andreas Wendker.

[ Cheers and applause ]

ANDREAS WENDKER: Good afternoon.

Welcome to WWDC 2015.

This is another exciting year to be an Apple developer.

With the addition of the new watchOS SDK, there are now three major platforms for your apps.

Our approach with these platforms is integrating great products with unique user experiences, while leveraging the same underlying OS technologies, APIs, and programming concepts so it’s easy to move between these platforms to share code and make these apps work together across the platforms.

But we are preserving a unique flavor for each of them.

The SDKs with these three platforms are going to ship later this year with a new version of our Xcode IDE.

It’s going to be version 7.

It contains a number of great new features.

I just want to highlight one here at the beginning of the session, that is that we are going to allow anyone with an Apple ID to download Xcode and run their apps on their own devices.

[ Applause ]

If you really want to make becoming a developer a lot more approachable, especially for our younger students, so this will work with all three platforms.

Once you have worked on your app and you want to deploy it, we have another bit of good news for you.

As of today, a single paid developer program membership will be enough to deploy your apps in all our App Stores.

You don’t have to sign up and pay for multiple memberships anymore.

[ Applause ]

Of course, with that single paid membership, you get many additional benefits, like access to pre-release software, or our powerful store analytics and crash reporting tools.

Now, let me talk about software updates a little bit.

The iOS adoption curve is, of course, the envy of the entire industry, and it’s a huge advantage to U.S. developers because you don’t have to deal with the same fragmentation that you find in other app ecosystems.

However, with iOS 8, we found that a number of users had a difficult time upgrading because of the large amount of free space required for the installation.

We’ve been bringing that number down with our subsequent iOS 8 updates, and we are continuing to do so with iOS 9.

As you already heard this morning in the keynote, iOS 9 will only require 1.3 gigabytes of free space.

We are also changing the way software updates are presented to the user.

Users will be given a choice to install right now or later at night, when they might not need access to their devices.

So we think this is going to keep pushing users to update quickly and allow you to focus your energies on the latest version of iOS.

Now, we are also working on a number of technologies that we call App Thinning that will enable you to return some space to your users.

App Thinning consists of three technologies: App Slicing, On Demand Resources, and Bitcode.

Let’s go over these one after the other, starting with Slicing.

Developing an app for iOS actually means developing multiple variants of the same app and then packaging them all together into a single app bundle.

So if you look inside an app, you find a number of redundant components to cover the full breadth of Apple devices.

There are binaries for 32- and 64-bit processor architectures; images for different screen sizes and resolutions; and resources like shaders, potentially written in different languages for the various kinds of GPUs.

But to run an app on each given kind of device, you only need a single slice of these components.

So to make the most of the available storage space, we are going to strip away all the components that are not actually needed on the device the app is running on.

So you would still develop and submit the same universal app that you are used to, but the store will only deliver a thin variant of the app to the user at install time.

[ Applause ]

So this leads to quite impressive size reductions.

As you can see on the chart, typical apps will get savings in the range of 20 percent to 40 percent.

And leveraging App Slicing is particularly interesting for apps like games that often bump up against the download size limit for cellular app installs.

The thin variants will have a much easier time staying under that size limit, so you can now pack more device-specific resources into your apps and provide the users with a more refined experience.

And best of all, assuming you are using Xcode’s asset catalogues, you don’t have to change a single thing in your projects.

The Store will simply do this automatically for you the next time you submit your app.

If you’re using additional custom data formats, we ask that you opt into slicing by creating an Asset Catalog and using the new asset categories we added to declare for what kind of devices your resources are needed.

Now, some apps benefit from using even more resources, through they usually don’t need them all at the same time.

For those kind of situations, we are introducing On Demand Resources, or in short, ODR.

With ODR, the store will separate your resources into the appropriate device slices and then host them on Apple servers.

But downloading gets delayed until your app explicitly requests them, and then later iOS will reclaim that space when your app doesn’t need the resources anymore.

Typical candidates for ODR are level-based games, but many other apps can benefit from ODR as well.

For example, you might want to consider offloading tutorials into ODR assets so that they only get downloaded when the user actually wants to watch them.

Using ODR requires a little more planning on your part, but the important point is that you can use more than 4 gigabytes of resources, just not all at the same time.

So App Slicing and On Demand Resources are going to help greatly with the storage space for an app.

But we are also working on a new technology that is a little more forward-looking.

We are introducing Bitcode into the App Store submission process.

Bitcode is an LLVM Intermediate Representation of your binary, which allows the store to reoptimize your apps for each kind of device before they get delivered to the user.

You develop and debug the app the normal way in Xcode, but the store will be in a position to constantly reapply the latest compiler optimizations we have been working on so that your apps run great on all kinds of devices.

It also future-proofs your apps because it will allow the store to let your apps take advantage of new processor capabilities we might be adding in the future, and all that without you having to resubmit your app to the store.

[ Applause ]

So generating Bitcode is controlled with a simple setting in Xcode.

It will be mandatory for all watchOS apps right from the beginning, and for iOS it will be optional for now, though we feel that the benefits are so strong that we are you are going to opt all your projects into it by default.

So that’s App Thinning.

It consists of App Slicing, On Demand Resources, and Bitcode.

These will greatly optimize your apps, especially for storage space.

Now let’s take another quick look at the binaries in your apps.

The store currently requires you to submit both 32- and 64-bit versions of your apps.

Over the last few years, we’ve seen an explosion in processor capabilities on our iOS devices.

For our CPUs and even more so for our GPUs, and all that especially since we introduced our 64-bit processor architectures.

So this category of apps that are really only possible when they target these 64-bit processors 64-bit architectures.

So going forward, we will allow you to submit 64-bit only apps to the store, starting with iOS 9 later this year.

[ Applause ]

So next we want to talk about the watchOS SDK, and to talk more about that, I am going to hand it over to my colleague, Eliza Block.

[ Applause ]

ELIZA BLOCK: Thank you, Andreas.

We’re delighted by the reception that the Watch has gotten in the developer community.

There are already thousands of your WatchKit applications available in the Store, and today I am excited to get to tell you a little more about some of the great new features we’re introducing in the watchOS 2 SDK.

First, let’s talk about the architecture of an existing watchOS application.

You have a user interface, which you’ve constructed as a storyboard, and this is installed on the watch.

Powering this user interface is your app extension, which runs on the iPhone.

In watchOS 2, we are making a significant change to this architecture.

The user interface remains the same, but the extension moves over to run natively on the Watch, and this has a number of advantages.

Because interactions with your Watch application no longer require a round-trip to the iPhone, your users are going to notice a substantial improvement to the speed and responsiveness of your application.

And of course, for the same reason, your Watch application can now work even when the iPhone isn’t present.

And since the extension is running natively on the Watch, it gets access to the Watch hardware, which opens up tons of possibilities.

Okay. So now that you’re writing a native watchOS application, what APIs are you going to use?

The good news is it’s many of the same frameworks that you’re already familiar with from developing for iOS.

There’s one important addition.

The watch connectivity framework allows your extension to talk to the iPhone app, which, of course, is now running on a different device.

In addition, your extension can talk directly to web services using the NSURLSession API.

And this works even when your iPhone is out of range.

[ Applause ]

There are three ways of surfacing your data in an existing watchOS application Glances, Notifications, and the app itself.

But wouldn’t it be cool if you could see your app’s content just by raising your wrist without even having to touch the screen?

In watchOS 2, this is possible because now you can create a Complication for the Watch face.

We designed the Watch for brief interactions, and the quickest way to see your content is through Complications, Glances, and Notifications.

So let’s take a closer look at these.

Despite their name, Complications are actually quite simple.

They’re snippets of information that appear on your Watch face alongside the time.

So if we take these Watch faces and we remove the time, everything left is a Complication.

Now, the ones you see here are all built right into the OS, but now you can create your own.

Now, you might not all be aware of this, but the fictional San Francisco Soccer Club is currently holding its 2015 tournament.

So we could create a Complication for the modular face to show what match is coming up next in the tournament.

If I were to install this on my Watch, I am going to see it every time I raise my wrist right away, and that’s great, but for it to be a good experience, it’s important that as time goes by and the content that the Complication needs to show changes, this has already happened by the time the screen turns on.

And to make that possible, we’re going to be collecting the data for your Complications in the form of a timeline.

That way, as I glance at my watch throughout the day, the Complication will always be showing the information that makes sense at that moment.

So creating a timeline for the schedule of a fictional soccer tournament is pretty easy.

For one thing, I made up the schedule, and for a second thing, it’s pretty unlikely to ever change.

But that might not be the case for your Complications.

As things change in the world, you’re likely to need to update your timelines accordingly.

And we’ve provided a couple ways to go about this.

The first is Scheduled Updates.

Scheduled updates are perfect for Complications whose data changes on a predictable schedule, like a weather forecast.

If you’ve written a weather forecast Complication, you may know that your server has access to more accurate weather data maybe once an hour.

So when you load your timeline, you can tell us a good time to next wake you up to refresh your data.

Your extension will be given the opportunity to run in the background.

You can hit your server, pull down new forecast data, and reload your timeline.

But not all data can be refreshed at predictable intervals.

Suppose I wanted to add live scores to my soccer Complication.

It’s not going to be good enough for me to hit my server every hour or so or even every 15 minutes to pull down the current score because when a goal is scored, that needs to show up right away in my Complication.

To support this kind of case, we’re introducing a new kind of high-priority push notification.

If you are providing data that is needed for a Complication, you can send this push notification to your iPhone, and it will be delivered immediately to your extension on the Watch.

So timelines are great for making sure the content in your Complications is always up to date, but they also serve an important additional purpose.

If you went to the Keynote this morning, you will have heard about the Time Travel feature that we are introducing in watchOS 2, which allows you to turn the Digital Crown to peek forwards and backwards and see what your Complications will be showing at different times of day.

For example, here it looks like the Marina Noe Valley game has already ended and I missed it because I was on stage talking about WatchKit.

Luckily, I can Time Travel backwards to see what happened.

It looks like Noe Valley won in stoppage time.

That must have been really exciting.

So that’s Complications.

They are a really quick way to access data that’s important.

Small amounts of data.

But what if you wanted to see a little more detail?

That’s when you might use a Glance.

You get to your Glances by swiping up on the clock face, and Glances give you the opportunity to display an entire screen full of data to your user.

Here, for example, my soccer club Glance is showing me the standings for Group A in my tournament.

Complications and Glances allow me to access your app’s data on my own schedule.

But sometimes you need to get information in front of your user right when it matters, and for that you would use a Notification.

Notifications on the Watch are incredibly powerful.

You can take full advantage of the screen to show an entirely custom UI.

For example, here I am being alerted that it’s my last chance to pick the winners in this afternoon’s SoMa Cow Hollow game.

In addition to the custom UI, you can also provide custom actions.

Pressing either of these buttons would register my pick with the application.

But that’s not all.

Because in watchOS 2, we are introducing Notification in-line text reply, so I can add a button to my Notification allowing me to compose a message to send.

I am a really big SoMa fan, and I want the world to know it, so I am going to choose that option.

When I pick the Reply option, I am given the Quick Reply sheet, and I can even use dictation to compose a message, all from inside the Notification without switching context.

Thank you.

You are very kind.

[ Applause ]

So that’s Notifications.

So Complications, Glances, and Notifications are so easy to access, they are probably the way that your users will most commonly interact with your app’s content.

But sometimes you have a little more time and you want a more interactive, immersive experience, and that’s when you would launch the full application.

And there is so much you can do with applications in watchOS 2.

To give one example here, now that you have access to input from the Digital Crown, you can use it to scroll quickly through the groups of your tournament to quickly find your favorite team and get more information.

But that’s just one of tons of new opportunities that you have with watchOS 2.

You can also add animations to your UI now.

You can access the Taptic Engine.

You can do audio recording right from the Watch.

And you can embed audio and video playback.

You can make a phone call from your application, and you get access live access to the data from the health sensors and the Accelerometer.

We are so excited to see what you guys create with all of these possibilities, and now to show you how easy it is to bring your app to watchOS 2, I’d like to invite up Josh Shaffer.

[ Applause ]

JOSH SHAFFER: Thanks, Eliza.

So I am really excited to show you how easy it can be to take your existing watchOS 1 app and upgrade it to watchOS 2 to run natively on the Watch.

If you have downloaded the WWDC app this year, you may have noticed that it now includes a WatchKit extension to install and run on your Watch.

So what we are going to do is take a look at how we can update that application and use it as a native watchOS 2 application, and we’ll add a few features to it using some of the things Eliza showed us.

So over here, you can see that I have the Xcode project open for the WWDC app, and Xcode has noticed that I have a watchOS 1 extension and is offering to upgrade it for me to a watchOS 2 extension.

I am just going to click here and let it perform those changes.

It’s going to take all my existing targets and update them, leaving my source code and storyboards in place so I can reuse all of what I’ve already done.

We can go over here and check out our controller context, and what we have to do first of all is make a few code changes to the way we access our data.

Now that we are running on a different device, we need to be able to move the data from our iPhone over to the Watch, and we can use the new Watch Connectivity frameworks to do that.

So I will start adding the Watch Connectivity framework, and we will replace the open parent application call, which is what I used in the watchOS 1 version, with the new Watch Connectivity code that will ask for the data to be copied from the phone to the Watch.

With that changed, now we can add some additional functionality, and one really cool thing to take advantage of is the ability to interact with the Digital Crown.

Now, the way that you interact with the Digital Crown in a watchOS app is using a new interface picker control.

We will drag that out in our Storyboard in just a minute, but first we are going to add some code to hook up to it.

We will make some references to be able to hook up to the Storyboard object once we drag it out, and you add the items that you want to choose from the picker programmatically.

So we are going to loop through all of the session tracks that exist throughout the week and add an entry to our picker for each track.

This way we will be able to sort the list of sessions and view just the ones for the track that we are interested in.

Then finally, we have to add an IBAction that will hook up to our control that will get called every time it changes.

So we will add that code there.

Now we can switch over to our Storyboard and figure out where to put this in our app.

So I will come over here and find the new picker control, and we will drag it out and put it in our session controller.

The session controller is the view that displays all the list of sessions, so by adding at the top, it will provide a way to filter that session list.

Just make it a little shorter so it’s not quite so tall.

Now, the picker control is really flexible, and there’s three different appearances that you can choose list, stack, and sequence that you can learn about throughout the week.

The list one is a really good choice for what we are doing here, so we are going to leave that.

And we are going to turn on a focus outline so that we can make sure the user can tell when they turn the Digital Crown what’s going to happen.

So with those changes made, we’ll then hook up our picker to the code that we pasted in a minute ago.

We’ll drag out the connection to our IBOutlet and drag a connection into our IBAction so that it gets called every time the picker changes.

And that’s it.

So with that, we are going to hit Build and Run, and we can switch over to the Watch simulator and see how this works.

The great thing about the watch simulator in watchOS 2 is that it’s a full simulator in the watchOS, so you can run your entire watch app side by side with your iPhone app, debug at the same time, and test the connectivity code between them.

[ Applause ]

So now we can jump over here, and you can see in our session list, we can scroll through the list of sessions, and we can also filter based on what track we want to see, and it all updates as we change it.

Now, one last feature that I would really like to add is the ability to display the sessions right on our Watch face, and to do that, we are going to add a Complication to our app.

Now, I wrote most of the code for this earlier, so I am just going to drag the file out and add it into my project right here.

The one thing that I haven’t done yet is added the code that will iterate through and build the timeline.

Now, Eliza mentioned that we are going to be providing the data in the form of a timeline, so we want to iterate through all of the sessions that I favorited so that on the Watch face I can see my favorite sessions all week long.

Now, to save us some time, I built and installed this on a Watch earlier today, so I am just going to switch over here and take a look at how we can add that to our Watch face.

So right at the beginning, I had a Watch face configured with Calendar in the middle, but because I’ve got all this on the WWDC app, we will just switch over here and customize it, and then we can scroll down to the bottom of the list of Complications, and you will see that now the WWDC app has an entry all the way at the bottom.

We can click there to turn that on and go back and view our session list.

Now, obviously, the platform State of the Union is the session I favorited for right now, but with no additional work, we can Time Travel forward through that session list and see that up next is the Apple Design Awards, which I definitely don’t want to miss, and then looking forward to tomorrow we’ve got the Intro to WatchKit for watchOS 2 in the morning, which is definitely something I want to see.

So that’s how easy it can be to update your existing Watch app to run natively on the watch as a watchOS 2 application, and add some support for some great new features.

So next up [ Applause ]

Next up, Sebastien Marineau-Mes is going to tell us about some great new foundation technologies.


That was great.

Let me now talk about a number of foundation technologies that apply across our platforms, and I am going to start with the first one that’s in the theme of performance, and it is compression.

We are going to make compression exciting.

[Laughter] So Apple’s always delivered a number of compression algorithms as part of our core frameworks, LZ4, which is optimized for speed; LZMA, which has high compression, and zlib, which many of you use, which hits the sweet spot between compression and speed.

We set out to build a better compression algorithm that improves on zlib.

As you may know, zlib is over 20 years old.

It was built when processor architectures were very different, so we thought let’s build something optimized for today’s processors and microarchitectures.

Now, we call this new algorithm Lempel Ziv Finite State Entropy.

You may know of them.

They are also a great Indie band.

You should go check them on Connect.

To avoid any confusion, we are going to simply call this LZFSE.

[Laughter] So now LZFSE thank you improves on zlib on compression, but more importantly, it actually is three times faster than zlib.

This is a great improvement in terms of making your apps more responsive.

On top of that, it helps with battery life.

It gives you a 60 percent reduction in energy use with the same compression algorithm, it also gives you a 3x speed improvement.

Finally, we’ve made it super easy to adopt.

We’ve added it to our standard framework.

All you need to do is switch your algorithm to use LZFSE.

We do all the hard work for you, and you and your users can reap the benefits.

So this is LZFSE, really redefining the sweet spot in mainstream compression.

Next up, I want to talk about battery life.

Craig mentioned battery life this morning.

Now, in our industry, when we test battery life, we often do so using repetitive tests.

For example, we’ll do web browsing over and over again as our test.

And we know that in the real world, the way that we use our devices is very different from these synthetic lab tests; right?

We take the device in and out of our pocket, we receive notifications and messages, we may use a variety of applications, we may go in and out of network coverage.

So what we’ve done in iOS 9 is really focused on a broad set of optimizations that apply across all the most popular apps on our platform.

That’s step one.

Step two, we focused on a number of infrastructure improvements.

For example, the algorithms that drive the backlight intensity, facedown detection so that if you receive a notification and are not able to see it because the phone is face down, we will simply not light up the screen.

A number of other optimizations in how the system, for example, sleeps, and optimizing its lowest power states.

All together, these optimizations add up to one hour of extra use sorry of typical use for average users.

Great improvement.

Craig also talked about Low Power Mode this morning.

Well, we’ve made it easy to turn on Low Power Mode.

Simply go in Settings and turn it on.

It activates a number of internal levers.

For example, preventing the CPU and GPU from entering the highest performance, but most power-hungry, states; preventing applications from doing too much work in the background; preventing them from doing unnecessary network activity; and finally, turning off a number of power-hungry animations.

And altogether, Low Power Mode, when you start from a full charge, will give you an extra three hours of usage.

So again, great improvement.

Let me now turn to protecting customer data.

So from the start, iOS has really focused on protecting customer data, and with iOS 9, we are taking a number of important steps to protect data, both on device and in the cloud.

So let’s start with the cloud.

Of course, use your Apple ID and use that to access a number of services in the cloud, your photos.

You might have documents that are on iCloud, your purchases.

Use it for messaging with iMessage and FaceTime.

And the biggest threat to your data in the cloud is that somebody gains access to your password, either because you’ve shared that password on another website that’s been compromised or perhaps an attacker, through a phishing attack, is able to gain that password.

The solution to this is two-factor authentication, and we are making it easy for everyone to adopt.

So let me show you how it works.

So with iOS 9, when you get a new device and you want to sign it up to your existing iCloud account, you will be, of course, prompted for your password, but on top of that, you will have to enter in a verification code.

Simultaneously, we send a notification to your existing, trusted devices, and that notification includes the location of this new device that’s signing in.

So if that’s you that’s signing in, you can simply enter the verification code.

But if it’s an attacker that’s trying to sign in to your account, you can stop them in their tracks.

That’s the cloud side.

Let’s now talk about the device where the passcode is your last line of defense protecting your personal data.

So historically, we’ve kept that passcode at four digits because you had to enter it every time you wanted to use your phone.

But of course, these days we don’t enter the passcode very often.

We use our fingerprint with Touch ID to sign in seamlessly into our phones.

So now with iOS 9, we are able to extend the length of the passcode from four digits to six, increasing the strength of the passcode by a factor of a hundred, without compromising ease of use for our devices.

Now that we’ve talked about the device, let me introduce a new technology on all of our platforms, which we call App Transport Security.

App Transport Security is really all about securing data as it transits over the Internet from a device or from your applications to your backend servers.

So of course we want that information as it transits over the network to be secure.

But it turns out that today it’s actually hard for you to do that.

It’s hard to get it right.

You can’t use HTTP.

You have to use a secure protocol, but then you have to worry about protocol versions, downgrade attacks, encryption, vulnerabilities.

Keeping it straight and doing what is best practice is difficult.

With App Transport Security, we are building this right in to our core frameworks.

And so now when you use our standard networking frameworks, we will enforce a secure best practice connection between your application and your backend servers.

Today that best practice is TLS 1.2 and Forward secrecy, but as the standards and the state of the art evolves, of course the framework will implement that new state of the art and will enforce it for you automatically.

[ Applause ]

So where do you go from here?

If you use out standard networking frameworks, for example, NSURLSession, it is built right in.

It is on by default with iOS 9, so when you download the developer beta which I am sure many of you already did and you recompile your app, we will enforce best practices secure connections to your backend.

It is possible that you haven’t updated your backend yet to support best practices or perhaps you are using a third-party library.

If that’s the case, we have provided an exception mechanism to give you time to update your backends through your info.pls.

So that’s App Transport Security.

Let me now switch and talk about the Mac and introduce a new technology which we call system integrity protection.

Now, on the Mac, user accounts typically have administrative privilege, and that administrative privilege is really equal to kernel-level access, and it makes it difficult to protect the integrity of the system.

With system integrity protection, we break that equivalency and administrative-level access is no longer sufficient in order to do a number of operations in the system.

Let me show you some examples of this.

So for example, if you have administrative-level privilege, you cannot modify system files.

You can’t install to system locations.

You can’t attach to system processes and, for example, introspect the memory or change the control flow.

So with the beta that’s out today, we encourage you to download it, test your apps to make sure that you are able to basically adhere to all of these rules.

Now, you might also be wondering how will this affect the way that I develop?

Well, the good news is we’ve updated Xcode and the development tool chain, and for the vast majority of you, you will see no difference.

For those that have a specific development requirement for example, you are developing kernel extensions on OS X we do provide a utility that’s part of the recovery partition that allows you to disable system integrity protection.

Which takes me now to the last technology that I want to cover, one that underpins the modern Internet, IPv6.

Now, why is IPv6 important?

We’ve all heard about the shortage of IPv4 addresses.

Well, guess what.

It’s finally here.

In fact, in China and Asia, they ran out of IPv4 addresses back in 2011, and in the U.S., we will be running out in the next couple of months.

What it means in practice is that a number of carriers are now deploying IPv6-only networks.

And if your application doesn’t work properly with IPv6, it will simply not function on those carriers, those networks, and for those customers.

So really important to support IPv6.

Now, the good news is we’ve had IPv6 support as part of the platform for a long time, well over a decade on Mac.

It is field proven, and many of you that are using our standard networking frameworks are making use of IPv6.

But we want to go further.

We want to make sure that every one of you is building an app that works on IPv6 networks, so we’ve got a simple recipe for you.

Use the standard networking frameworks.

We’ve mentioned NSULSession.

It really takes care of extracting the complexities of the network.

Avoid the use of IPv4-specific APIs, many of which were developed before IPv6 was even conceived.

And finally, don’t hard-code addresses.

Once you follow the recipe, you might say, how do I test that it works properly?

Because not all of us have access to an IPv6 network.

The good news is again you all use Macs for development, and we are turning your Mac into an IPv6 network emulator with the latest beta.

So all you need to do is set up through a new connection sharing mode a personal hot spot, you check the IPv6 only box, then you can connect your test device, test your applications, and make sure that they work on IPv6.

It’s as easy as that.

[ Applause ]

So finally, because IPv6 support is so critical to ensuring your applications work across the world for every customer, we are making it an App Store submission requirement starting with iOS 9.

So that takes me to the end of my foundation technologies.

Let me now hand it over to my fellow Canadian, Toby Paterson, who will be talking about higher-level APIs.


[ Applause ]

TOBY PATERSON: Well, I know many of you have come from all over the world to be here today, some of you from even as far away as Canada.

[Laughter] Well, your customers are just as diverse as you are.

Here is an interesting fact you may not have realized about them.

This chart shows the worldwide App Store sales broken down by country.

One of the interesting things it tells you is that if your app is not localized properly, you are not going to reach potentially the vast majority of your customers out there.

Now, the good news is it’s not hard for you to do.

We have an internationalization and localization guide that tells you everything you need to know, and of course, we’ve got some great support in our frameworks and SDKs for this.

We have formatters for dates and numbers.

And this year we are introducing a new formatter to help you display people’s names properly.

Now, names are a tricky thing.

Everybody has one, of course, but different cultures represent them differently and have different conventions around their use.

In English, we typically write a name as first, middle, and last.

Chinese, on the other hand, leads with the family name and doesn’t even have the notion of a middle name at all.

The NS person name component formatter which I promise looks better in code than it sounds when you say it out loud [laughter and applause] thank you.

This takes care of all of the details for you.

And it even has some cultural smarts.

So if you ask it for the short version of a person’s name, it knows when it might be inappropriate to use that person’s first name all by itself.

Now, we have some really exciting news for Arabic and Hebrew speakers out there.

We’ve had rudimentary support welcome [laughter] we’ve had some rudimentary support for these languages for years.

On iOS it was limited strictly to text; whereas, the Mac had some more sophisticated layout of your UI elements.

Well, I am really pleased to tell you that in iOS 9, we have full mirrored UI support for right-to-left languages.

[ Applause ]

Here’s what it looks like in Springboard.

Now, we haven’t just reversed some of the UI elements on screen here.

This is a much deeper conversion, where we’ve reversed the flow between your view controllers, your user interactions, and the system gestures.

This is what the UI would look like if it had been designed by a native Hebrew or Arabic speaker.

Let’s take a look at what this looks like on the phone.

I’d like to bring Sara Radi up on stage to give you a quick tour.

[ Applause ]

SARA RADI: So let’s take a look at the new right-to-left languages support on iOS 9.

So here I am running my iPhone in Arabic, and starting from the lock screen, I slide from right to left to unlock my device.

Here, my app icons are also laid out from right to left, and my page view flows naturally from the first page to the second.

So the whole system and all our native apps fully support user interface mirroring or running in right-to-left localizations.

So let’s take a look at the Photos app, for example.

Here, all of my photos are laid out the way I expect them to be.

The entire UI feels just right from the navigation bar at the top to the bar at the bottom.

Now let me show you Mail.

In addition to the standard layout, navigation, gestures, and animations also flow as expected.

So hitting Mail to trigger the quick message actions, I am going to swipe from the left edge of the screen to the right.

And that compliments the layout flow of the table view cell.

Also, the navigation gesture works as expected from the right edge of the screen.

Now rotating my device to landscape reveals the Message view.

As a native speaker, this feels so natural to me since all my emails are on the right side and the details are on the left.

And since we’ve built all of this into UIKit, your apps will get the exact same behavior for free for running in right-to-left localizations.

Now let me show you a third-party app from the App Store.

And just in case you are wondering, we didn’t choose Kayak because the name is a palindrome and you can also read this from right to left.

[ Applause ]

So since it is using autolayouts and our new APIs with very minimal work, they just add translations, and their app just works as expected.

So after WWDC, I am planning to go on vacation, so I am going to explore some views.

So the first slider lets me pick the number of days of my vacation, so I am going to drag that slider to the right side to decrease the number of days.

Since they are using a stock UI slider, they got this behavior for free.

I also prefer nonstop flight only, so I am going to turn on that switch at the bottom, and as you can see, it also flows from right to left.

So the entire UI feels just right and feels very intuitive for natives of these languages.

And that’s how easy it can be to add right-to-left localizations to your apps on iOS 9.

Thank you so much.

[ Applause ]

Now back to Toby.

TOBY PATERSON: Thank you, Sara.

Well, as Sara mentioned, we have full support for this in UIKit and our other system frameworks.

So you can get a lot of this just for free in your own applications.

If you have custom views or gesture recognizers, you do need to think about what they mean in a right-to-left language.

We have this API here, which I am not even going to try and say out loud, to tell you which way the UI is flowing so you can make the appropriate decision for your UI.

Next I’d like to talk about links, and by these I mean the kind of links that you get from a friend in a message or an email.

You tap on it, and invariably it opens up in a web page.

Well, this is exactly what you want on the Mac because Safari is a natural home for web applications.

iOS, on the other hand, is all about the native app.

Wouldn’t it be great if that same link when you tapped on it could open up right in the associated application?

Well, that’s exactly what we’ve made possible in iOS 9.

[ Applause ]

This kind of universal app linking that takes you to the the same link that can take you to the natural destination for the platform that you are on.

Safari on the Mac or native apps on iOS.

Here’s how it works.

You host a file on your website that lists the kind of URLs that your application can handle natively.

When the user taps on one of those URLs, we wrap it up in an NSUserActivity and hand it off to your application the same way that Handoff does, and that’s it.

It’s that easy.

Now I’d like to talk about some of the new things in iOS 9, starting with HomeKit.

So this morning we announced a number of great new features for the HomeKit platform, and of course, we’ve been continuing to enhance the HomeKit framework.

Certain classes of accessories can now post push notifications right to your devices, doors, windows, and alarm systems.

And we have a new system UI for managing who you are sharing your HomeKit network with.

This is available in Settings, and we also have an API so you can bring it up right in your own application.

But I think the really interesting things are event triggers.

These let you set up “if this, then that” sequence of actions so that you can turn the lights off when you activate the alarm system or, more importantly, turn the coffee machine on first thing in the morning.

I think these are going to be really powerful.

Now we have some new things in Search that I think you are probably all interested in.

The big news for iOS Search, of course, is that now it can search your application content.

We can show rich results, and when the user taps on them, take them straight to the appropriate spot in your application.

There are three ways that you can index your application data.

Core Spotlight lets you explicitly index all of the application content that you have.

Now, some of that data may only be valid for a limited time, and so you can optionally provide an app indexing extension, which Core Spotlight will call out to at the appropriate time to make sure your indices are up to date.

Another way of indexing data is via NSUserActivity.

This lets you index things that the user is seeing or doing in your application so that they can easily find them and get back to your application.

And finally, if your app is mirroring content from your website, you can mark up that website content so that our web crawler will find it, index it, and provide results when searching locally on the device.

Now, you saw this morning that when you swipe into the new Search UI, we already have a series of suggestions ready and waiting for you without you having to type anything.

Well, Siri can even suggest results from deep within your application.

If you are using NSUserActivity to make a note of what the user is doing in your application, Siri can take that into account when coming up with these proactive suggestions.

You’ve probably noticed we are building a lot on top of NSUserActivity here.

There’s Handoff, of course, universal links, Search, Suggestions.

I think this is a great example of how we are building on our existing foundations to make your apps and the whole ecosystem much richer.

Now, the big news this morning, of course, was Multitasking, and we are really excited to bring this to you today.

There’s slideover and split view, which lets you pin apps side by side so you can work on them at the same time.

I could almost hear many of you wondering out loud this morning, oh, gosh, what do I have to do on my application now to take advantage of this?

Well, the answer hopefully is not very much at all.

You may remember last year we introduced this notion of Adaptive UI, which is a really simple concept.

It says that instead of designing your UI for fixed screen sizes, instead pick a layout and then adapt it to changes in your window bounds.

And we provide a number of tools to help you with this.

There’s Dynamic Type for laying out your text and autolayout constraints for adapting your UI to changes in bounds.

Now, we recognize, of course, that a single layout is not going to make sense across all these different screen sizes, so we introduced this abstraction called Size Classes, which try to allow you to choose the appropriate layout without having to resort to a device-specific check.

This works by broadly categorizing sizes, screen sizes, into two buckets, regular and compact.

You can see here that the iPad has a regular horizontal size class, and the iPhone has a compact horizontal size class.

We are using these exact same mechanisms for the Multitasking UI.

You can see here in the slide overview it has a compact horizontal size class.

And when I tap to pin those two apps in split view, we adjust the bounds of the main app there on the left but keep it with a regular horizontal size class.

Now, the interesting thing, when you resize that split view to 50/50, is not only do we adjust the window bounds there, but we also change the size class of the main app on the left from regular to compact.

We found in our own applications that this is the best layout for these window dimensions.

So if you have already adopted Adaptive UI in your app, there’s actually very little left for you to do.

You need to use a launch Storyboard, which is a more flexible replacement for default pings, and then declare that you support all orientations, and that’s it.

As Craig mentioned this morning, we have literally converted apps in a matter of minutes and had them up and running.

It’s really great.

Now, this morning we also announced Picture in Picture, which lets you continue to watch videos over top of your other applications.

And if you are a video application, it’s really straightforward for you to take advantage of this yourself.

The first thing you need to do, of course, is just support background media playback, and then you need to enable Picture in Picture support in your controller.

We have this built into our standard media playback controls, and we provide an AV Picture in Picture controller to give you very fine-grain control over entering Picture in Picture.

Next I’d like to talk about the Mac.

So OS X, of course, has had Multitasking, multiple windows forever, and this year we announced some new window management features that make that even more compelling.

The key element of this is the ability to tile two applications side by side in a split view, as you can see in this screenshot here.

Now, adoption of this is really easy.

Any resizable window can partake in this, and the key thing for you to do there is to make sure that your window lays out nicely in both narrow and wide geometries so that it can play along well with just about any other window in the system there.

And of course, we’ve added some new APIs and behaviors to AppKit to make this really easy for you, such as automatically hiding the sidebar in the NS split view controller.

Another key development on the Mac is the introduction of Force Touch, which we released a few months ago.

So Force Touch opens up a whole new dimension of Touch, with pressure sensitivity and haptic feedback on the Trackpad.

We use this in our own apps for things like text lookup, pressure-sensitive playback controls, and finer control over drawing, preview, and markup.

Naturally, we’ve added some API for you to take advantage of this in your own application.

You can set the pressure configuration for the kind of Force Touch that you want, configure the default haptic feedback on the Trackpad, and have pressure changed events delivered to your views and gesture recognizers.

We think Force Touch is a really exciting new capability for the Mac.

We can’t wait to see what you guys do with it in your own apps.

Now, I have some updates on iCloud for you.

Last year we launched iCloud Drive, and this is a great way for accessing all of your documents across your mobile devices, your computers, and the web.

Now, on iOS, apps provide a great in-app experience for managing your documents, and we think this simple app-centric model is great for many people out there.

But Mac users, of course, are accustomed to working with their documents directly in iCloud Drive.

So for those folks, we are adding a new application to iOS, the iCloud Drive app.

[ Applause ]

This is what it looks like.

You can see all of your documents in iCloud Drive organized just as you would expect.

You can preview many document types right in the application.

And naturally, you can open a document in its own application.

Now, prior to iOS 9, the only way of accomplishing this was to copy the document into the application container, and of course, that’s exactly what you don’t want for documents in iCloud Drive or any other document provider for that matter.

What you really want is to be able to edit those documents in place without moving or copying them.

So if you are a document-based application, here’s what you need to do to enable that.

You need to support file coordination since there may be multiple processes trying to access that document simultaneously.

And then you just add this key to your Info.plist and handle the application open callback.

And that’s it.

iCloud Drive is built on top of CloudKit as are, in fact, many of our iCloud services.

CloudKit is a public API, and when we launched it, we said it was going to be basically free for probably most of you.

Well, what does that actually mean in practice?

Here’s what you can store in iCloud Drive for free right from the get-go with no users.

As you add more users to your application, these grow commensurately up to these maximum data limits.

Now, we’ve built a CloudKit Dashboard for you that shows you all of the key metrics at a glance and includes a line below which everything is free for you.

We project that line a short ways out into the future so you can have some advance notice before crossing it.

And if you do cross that line, we’ve published a clear and simple pricing guide here.

You can also find tons of other interesting information about CloudKit here, such as the details of a new feature that we are launching this year, CloudKit Web Services.

This is basically a full-on CloudKit implementation for web apps.

Anything that you can do with the native CloudKit API you can now do via JSON.

We provide a JavaScript library that mimics as closely as possible the native CloudKit API.

And we provide a web login flow so that you can authenticate your users securely.

We think this is going to make it really easy to write a CloudKit-based web application to run alongside your iPad, iPhone, and Mac apps.

And that’s what I have for CloudKit.

Thank you.

[ Applause ]

Now I’d like to hand off to Chris Lattner, who is going to tell you something about Swift.

[ Cheers and applause ]


Thank you.

Thank you, Toby.

The response to Swift has been amazing, and it’s really actually hard to believe that it was unveiled just one year ago.

When we talked about Swift back then, we talked about some of its major features like performance, safety, and also the interactivity of Playgrounds.

But we want Swift to be pervasively available to everybody, and so we are open sourcing the compiler and the standard library and even kicking things off with the Linux board as well.

[ Applause ]

This project is going to be run under an OSI-approved permissive license starting later this year after we wrap up work on Swift 2.

We look forward to working even more closely with you all, incorporating both your ideas and your code contributions as well.

So there’s been a lot written about Swift, and we’re not going to go through all the quotes, but one of our favorite was when Swift topped the list of Stack Overflow’s most-loved programming languages.

And there’s a lot of reasons to love Swift; right?

One of which is it’s built to run your applications at top speed, and over the last year, performance of Swift code has grown tremendously as we’ve implemented new optimizations in the compiler.

And there is a ton new in Swift 2 with improvements across the board.

But let’s talk about a few of these, and maybe we will start with one of the most requested features, a new error handling model.

So error handling is a very well-known field.

There’s a lot of known approaches.

But all these approaches have very well-known problems as well.

So we weren’t satisfied with any of the existing approaches, so we came up with something we think will feel familiar but will solve these kinds of problems.

So let’s talk about it now.

It starts out simple.

A method or function in Swift can now be marked as being able to produce an error.

This greatly simplifies many common Cocoa APIs and also allows the compiler to verify that you are handling your errors properly.

Swift now has familiar catch syntax to handle errors and uses powerful pattern matching to enable you to express rich cache conditions.

Swift uses the try keyword but uses in a different way.

It uses it to mark calls to methods that can produce an error.

This defines the way an entire class of errors that happen when you have unanticipated control flow by making that control flow explicit in the code so you can reason about it.

Of course, it’s easy to throw an error, like you’d expect, and Swift is the perfect way to define your own categories and families of your own custom error conditions.

It works really great.

Now, we think that Swift error handling in Swift will be super familiar and will feel really natural, but it will also greatly increase the quality of Swift code across the board.

Let’s move on now and talk about our next big feature, availability checking.

So we introduce great new APIs all the time.

Often you want to adopt some of these new APIs to get new capabilities in your applications as well, but you can’t always drop support for the previous OS X or iOS release.

Now, this brings a challenge because some symbols you want to use are unconditionally available, where other symbols are conditionally available depending on the OS that your app is running on in the field.

With Swift 2, handling this is a breeze.

Now if you try to use a symbol that’s conditionally available without checking for it, the compiler will produce an error message, so you can’t mess it up.

And Xcode goes even farther by giving you several great ways to handle this.

The first is you can use the new “if available” statement to add a fine-grain check right where you need it.

[ Applause ]

But I think even better than that is that you can also mark an entire method or even an entire class as dependent on new OS features.

This eliminates the need to do fine-grain checks and directly expresses many common situations that you will have in your code.

It’s a great solution that works really well together.

Now, the error handling features and the availability checking features of Swift 2 are two great ways to make it easier to write correct code.

But, of course, we want your code to be beautiful as well.

So we’ve done several great changes in Swift 2 to help this.

Swift 1 introduced a number of really powerful global generic functions to do powerful things with algorithms.


Powerful and wonderful.

The problem is these were not always beautiful to use.

Let’s say I have a collection of numbers, I want to scale them up and drop some of the results.

With Swift 2, this is easy, but writing it requires rearranging a lot of code, and when you look at it, you have to read it inside out, which makes it difficult to reason about.

Swift 2 introduces a new language procedure called protocol extensions.

With protocol extensions, we can now recast these global functions as methods, the way they should have been all along.

This means that the new code that you write is beautiful.

It’s simple to write, and it’s simple to read.

[ Applause ]

Let’s talk about early exits.

It’s very common to want to exit a scope early for some reason, so maybe you have a method that takes a parameter that could be nil, and if it’s nil it doesn’t want to do anything.

Well, the “if let” statement gives you a very familiar, comfortable, and great way to check these conditions, but it requires you to indent all your code an extra level.

To solve this problem, we have introduced a new guard statement that allows you to check a condition and then bail out early.

It allows you to write nice straight-line code without the excess indentation.

[ Applause ]

Next, a huge component of how Swift feels is how well it works with Cocoa.

But with a plain Objective-C API, the Swift compiler has no idea whether or not a pointer is nullable or what the element types of a collection are.

We solve this by introducing new features subjective C, including the ability to express nullability for pointer types, and in Xcode 7, we introduced an entirely new first class generic system to Objective-C that allows you to express element types and many other things right in Objective-C.

[ Applause ]

Through the adoption of this and a whole bunch of other features, Cocoa is feeling quite swift.

Finally, let’s talk about Xcode.

Xcode 7 introduces a revamped Swift migrator.

It automatically will move your Swift 1 code to Swift 2 syntax, including adoption of the new error handling model.

As we continue to evolve the Swift language going forward, we will keep moving the migrator forward to match.

Next, let’s talk about header files.

But wait. Header files and not having header files are a huge feature of Swift; right?

The problem is, is that sometimes you actually do want to skim a bunch of code to understand what it does at a glance, and all the implementation details get in the way.

Well, Xcode has solved this problem by introducing a new Assistant editor, which gives you a header file-like view of an arbitrary Swift source file.

This gives you all the skimmability advantages of a header file without the maintenance burden of having to write, maintain, and edit it yourself.

That’s great.

[ Applause ]

Next, rich comments.

Xcode allows you to write beautiful Swift Playgrounds by writing rich comments directly into the Xcode editor.

These rich comments use the very popular Markdown syntax, and now we’ve brought in the syntax to documentation comments as well.

We’ve even added support for new features, like in-line images and links, giving you a great, consistent experience between Playgrounds and doc comments.

Finally [ Applause ]

finally, let’s talk about Playgrounds themselves.

Playgrounds are very important to us.

We know that they are a great way to experiment with API and with the Swift program language itself.

But we think that Playgrounds can be a great way for teaching and learning programming as well.

And so we have been adding a number of features to Playgrounds, including most recently support for multiple pages within a Playground.

To show you this and more, I’d like to invite up Max Drukman.

[ Applause ]

MAX DRUKMAN: Thanks, Chris.

Today I’d like to show you how easy it is to learn and to teach using Playgrounds in Xcode 7.

Let’s start off with a playground that I am developing to teach students about some of Apple’s great graphics APIs, and in this section, I am introducing them to a brand-new one, GameplayKit.

As you Playground says, GameplayKit is used to develop the mechanics that drive your gameplay.

We will see a little bit more of that in a minute.

This Playground uses a car chase visualization to demonstrate some of the GameplayKit capabilities, and I have created this Playground so that my students have the ability to edit the bad guy cars by tweaking these three Sprite Node variables.

Now, Playgrounds have always had the ability to show you results for each line of code.

This year, we added the ability to add them in line.

So now I can start by editing my Playground the way my students will, by changing the values to get a different look to the bad guy.

And you can see that as I make my edits, the changes are updated live.

Now that’s a bad guy car.

[Laughter] Down here, I am asking my students to write a little bit of code to put the pieces of the bad guy together into one Sprite.

So now as I write that code, it’s going to assemble my bad guy, and now I am ready to go.

That’s it for the setup.

Now it’s time to move on to the main event, which is making stuff move.

Which I have on another page.

Now, pages are a great way to divide up your Playgrounds into subtopics, kind of like a book.

You can navigate to pages using the handy navigation links at the bottom of the page.

You can use the jump bar.

And of course, the navigator.

Here are all the pages in my Playground.

Now, each page can have its own sources and resources, so you can factor your Playgrounds exactly the way you want.

Let’s check out the next page.

I am going to put away the navigator for now.

So this page talks about GameplayKit’s flocking behaviors.

Now, flocking is what’s going to take the bad guys and make them move as a unit.

Now, without any further ado, I am going to open Xcode’s timeline Assistant and let’s meet the flockers.

Okay. Here are all my evil but colorful bad guys tracing our intrepid truck-driving hero.

But you can see they are kind of driving all over the place.

They are a flocking disaster.

[Laughter] Let’s see if we can fix that.

So there are several parameters that you can adjust in order to govern the flocking behavior.

Here is an interesting one, cohesion, which governs how much the bad guys want to stick together.

I can just play around with that value and see immediately what the effect is.

That’s a little bit too much cohesion, so I think I will back that one off.

That’s a little better in terms of spacing, but they’re still not quite as goal-oriented as I’d like, so I am going to play with the seek parameter, and give that a different value and see if I can get a good flock going here.

All right.

Now they’re flocked.

So now I know what kind of range of values I want to tell my students to play with.

I am going to put away the assistant, and I am going to switch the editor to be in raw markup.

So now I can edit the prose of my Playground using familiar Markdown syntax.

And I can go back down here to the seek parameter, and I can add a little bit of extra prose to tell my students to play around with these values.

I can even come up here and add some branding in the form of an icon.

And now whenever anyone asks me if I know about flocking, I say I wrote the flocking book.


Thank you.

You are very kind.

[ Applause ]

These are just a few of the authoring features we’ve added to Playgrounds in Xcode 7.

Can’t wait to see what you build with them.

And now, to tell you about some more great Xcode features, I’d like to invite up Matthew Firlik.

[ Applause ]

MATTHEW FIRLIK: Thank you, Max.

Alongside the new releases of watchOS, iOS, and OS X, we are excited to bring you Xcode 7.

This new release includes features and support for all of our platforms, and the unique experience of each platform has been integrated into the tools to allow you to target and deliver your applications to all of our devices.

That unique experience is where I’d like to start today, with our design tool Interface Builder.

And first up is a new way for you to lay out your applications called stack view.

[ Applause ]

With stack view, you can design your applications using horizontal and vertical stacks.

You start by selecting an orientation and add your controls.

With each control that you add, stack view automatically adjusts sizing and positioning to give you just the right look.

Stack views nest beautifully, and they allow you to change orientations, so you can get the precise layout you want.

Now, we’ve built stack view on top of autolayout, and it handles all the constraints for you.

Zero constraints gives you control over alignment, spaces, and distribution in the stack.

And we’ve made stack view flexible as well, allowing you to play with your interface in Interface Builder and at runtime.

When you add controls into a stack view, you can reorder them to try out different layouts and, perhaps best of all, when you hide views at runtime, stack view automatically adjusts.

[ Applause ]

So stack view, a new way in Interface Builder for you to get the precise layout you want.

Next up are Storyboard references.

With storyboards, you can create and design scenes and connect them together with zero code to make complete user interfaces.

Now, knowing that your scenes and the interfaces evolve and become a little bit more complex, we wanted to make this easy to manage.

Storyboard references allow you to take a section of your interface and move it to a separate storyboard file, continuing to make connections to and from the scene.

[ Applause ]

With storyboard references, you can keep your interfaces focused and modular, the same way you do as your application code, and still easily design your application flow.

Now, equally important to the flow and layout of your application is the look.

And in this release, we are making Interface Builder something to see.

In an upcoming scene, you will see more of iOS and OS X’s visualizations brought right into Interface Builder.

Blurs and shadows will render inside of the canvas.

Vibrancy will become a standard part of the presentation.

And your own designable controls can present masking and shadows that compose with other views in your scene.

And the combination of these visualizations alongside our preview editors, which allow you to further refine the look and layout of your interface for specific devices, means that now, more than ever, you can design your interfaces right in Interface Builder and see what your users will on their devices.

[ Applause ]

Another aspect of building your applications is managing assets, and Xcode includes tools to work with the new On Demand Resource APIs.

With Xcode 7, you can tag assets and files throughout your project to be downloaded and used on demand.

Each tag represents a collection of resources, and we call these a pack.

And you can apply multiple tags to any individual resource, allowing you to use that resource in different situations.

To help you manage your tagged resources, Xcode includes an overview of your tags in the Project Editor.

Here you can easily add, remove, and rename tags.

You can change the pack contents.

And you can also configure various aspects of On Demand Resources, such as the download order and the download priority.

When you build your applications, your tagged assets are automatically assembled into resource packs.

You just need to tag your assets and you are ready to go.

When you deploy your applications, your On Demand Resources will be hosted by the App Store.

While you are debugging, Xcode will host your On Demand Resources on your Mac and stream them down to your devices on demand to simulate the Store download behavior.

And for those of you using Xcode’s continuous integration, your bots will build and host your On Demand Resources, so you can easily test your apps over the air amongst your team.

And finally, to help you track your packs, the debug gauges will show you the progress and status of all your On Demand Resources while you are debugging.

So the combination of all these tools makes it really easy for you to adopt On Demand Resources in your apps.

Xcode 7 also has some great new debugging and profiling tools to help you further enhance the quality of your applications.

Sebastien mentioned earlier our focus on battery life.

Getting the most out of your devices and getting through the day on a charge is very important.

So we are making it easy for you to track the energy use of your application.

The new Energy gauge for iOS will show you CPU usage, calls to networking APIs and location services, and also show you your apps’ transitions from foregrounds to background states.

And as with our other debugging gauges, you can gain insight into the behavior of your application and, when you need more information, dive into Instruments for all the details.

And speaking of which, Instruments has been updated to be better than ever.

The track view is now more compact and makes it easier to visualize your data with new fluid gestures.

We also have a number of updated instruments and new instruments for you to use, such as core location profiling and a new GPU system trace.

And in this release, we are integrating Clang’s Address Sanitizer into our debugging workflow.

Address Sanitizer is a memory error detection system for C-based languages.

When enabled, Xcode will monitor memory used by your application and can detect common issues, such as buffer overruns.

When these are detected, Xcode alerts you and provides essential details to help you diagnose and debug the problem.

And unlike other similar tools, Address Sanitizer is fast, so fast you can use it with all of your interactive applications.

But even with great debugging and profiling tools, sometimes sometimes bugs get out and will cause crashes for your users.

To help you get these fixed quickly, Xcode has integrated support for crash logs.

For your apps submitted to [ Applause ]

for your apps submitted to TestFlight and the App Store, Xcode will provide symbolicated crashes for each submission.

The integrated view will provide an updated list of the top crashes for your apps, showing you the backtrace, details about the crash, and charts of recent occurrences.

You can retitle the crashes for easier tracking, you can add notes to yourself, and you can mark the crash as resolved once you fix the issue.

Now, when reviewing a backtrace, you want to make it easy, so Xcode has this Open In Project button that will load the backtrace into the debug navigator in your project, giving you the very familiar workflow to navigate your source code for the crashing frames and find and fix the problem.

[ Applause ]

Earlier this spring, we brought you crash logs for your iOS applications.

Starting today with Xcode 7, you can get crash logs for your OS X applications, and crash logs for your watchOS applications will be coming soon.

And in this release, we will also be bringing you crash logs for all app extensions on all platforms, so you can track and improve those as well.

[ Applause ]

Now, another way to improve your applications is with testing.

And we all love testing; right?

I am going to try that again.

We all love testing; right?

[ Cheers and applause ]

That’s good because testing is an essential part to delivering a great application, and we have two big new additions for you this year.

The core of our testing solution is the XE test framework and our test navigator, which combine to provide great ways for you to design and organize your tests.

Building on top of these, Xcode has provided a number of ways for you to test your application.

You can test the correctness of your APIs and measure and track their performance over time.

You can evaluate synchronous and asynchronous behaviors in your code, and you can run your tests locally in Xcode and continuously on your integration bots.

And all of this combines together to make it a really powerful unit testing solution.

But we wanted to kick it up a notch.

So this year we are adding in user interface testing.

[ Applause ]

This is built on the same testing foundation, so now you can develop correctness and performance tests that work through your user interface to further extend your testing coverage.

And speaking of coverage, we are adding that as well.

[ Applause ]

Code coverage is a great new way for you to evaluate your testing progress and ensure you have all the tests you need.

So let’s take a quick look at a demonstration of the new UI testing and code coverage features in Xcode 7.

So here I have an application my team and I have been working on.

It’s a SpriteKit game for iOS and iOS X called Demo Bots.

As we have been developing the application, we have been working on tests, and you can see here in the test navigator we have quite a few of our tests.

It’s helpful to know which area of our code has not yet been tested either because we have not yet written the tests or because our current tests are not complete.

With Xcode 7, this is exactly the kind of insight you can gain about testing with code coverage.

Xcode collects coverage data each time you test, so I am going to jump to the testing report, and we’ll see there’s a new coverage section.

Inside of the coverage section, all of the files in my application are listed with an indicator on the right showing the percentage of coverage each file has.

Now, in this list, I’ve sorted it from highest to lowest coverage, so all this might look good.

As we scroll down, we see I have some work to do here still.

For each file in the list, you can disclose the contents and see the coverage of the methods and functions contained therein.

For the file I’ve disclosed, the first two functions have 100 percent coverage, but the other three have no coverage at all.

The report gives you a good high-level overview of the coverage for your project, but we wanted to bring this information directly to you.

So we’ll see here when I navigate to one of the source files, we bring the code coverage information right into the source editor.

The darkened areas are the parts of my code that have not yet been tested.

And this is great.

It allows me to see where I need to focus, and it also works great with our assistant editor because you can have your source code and the tests side by side, and as you continually test, get updated information.

If we go back to the report and look at the classes that were not covered, many of them are user interface classes.

They are view controllers and the like.

So to help me with that, I am going to create a user interface test.

I’ve already set up a user interface testing bundle and class, and so I will select that here.

The test I want to write is testing the options panel of our game.

Now, the new APIs and XE tests give you the ability to interact with user interface elements, accessing their properties and triggering actions, and it’s really easy to write.

But you know what’s actually easier than writing a UI test?

Recording one.

Let me show you how that works.

I am going to collapse the project navigator to give myself a little bit more space, and with my insertion point in the test I want to implement, I am going to click this little Record button down here at the bottom of the editor.

You will see when I do this, Xcode launches my application.

I want you to pay attention in the test in the upper left-hand corner.

The first thing I want to do is click the Options button in my application, and you can see that Xcode’s recording inserted the corresponding line of testing code.

[ Applause ]

You can see it’s pretty simple API.

It’s asking the application for the window and the button and telling it to click.

For the next piece of my test here, I want to fill out my player name here, so I will start typing in my standard gamer handle, and you will see here that two cool things are happening.

One, while I am typing in the text field, the editor is updating live to show me the values.

That’s pretty cool.

The other thing that happened is that UI recording noticed I was interacting with the same user interface element more than once, and it refactored my test to create a local variable for it.

This keeps the tests readable and means that refactoring and reusing this code later is really easy.

Let’s continue selecting some options here.

You’ll see that the test updates.

For the last element, I want to select a different robot here, so I will click this button and select the blue robot.

I will click Done, and go back and look at my test.

Just like that recording was easily able to track all of my actions.

Now a test should probably do more than just poke things.

We should probably validate some values here.

So before we click the Done button, let’s insert some code to check that I have the right settings.

I am going to insert a little bit of code here that gets that value of the active robot from the interface.

It’s in a text field.

I will compare it to the value I expected, blue bot.

With no further ado, let’s run our test.

I will stand back.

Here we see it’s updating the values, selecting the right robot, validating our test, and our test passes.

[ Applause ]

So recording makes writing UI tests really easy.

We can look at our tests inside of Xcode and run them locally, and we can also have them run on our continuous integration bots.

I have bots set up for both the iOS and the OS X version of the application, and if I select my iOS version and look at my tests, when I filter down to the failed ones, I see something interesting.

I have four user interface tests that are passing on an iPhone but they are failing on an iPad.

To make it easy to diagnose your test failures, each test run includes a transcript of all the testing actions that took place, and I can see these by expanding the test.

Here’s a list of all the actions that took place in this test that’s been running.

Because user interface tests are visual, though, we wanted to go a step further.

So each one of these actions includes a screenshot.

So if I look over at the passing iPhone test, I can get the screenshot for how the test starts, there’s the start of our game.

I can look partway through at the test.

Here we’ve typed a value into the text field with the keyboard up.

And I can also look at the very end of the test.

Here’s the state of the UI just before we click the Done button.

If I flip over and look at the tests on the iPad and jump all the way to that last action, we can see here’s our iPad UI with all the right settings, but there’s no Done button.

Clearly we misconfigured something in our UI, but the screenshot was able to help us narrow that down really quickly.

So code coverage showed me where to write tests.

UI testing and recording made it easy for me to get more coverage.

And the new testing reports helped me narrow in on my problems.

That’s just a little bit of code coverage and UI testing in Xcode 7.

[ Applause ]

User interface testing makes use of our systems’ accessibility APIs to interact with UI elements.

This means if you’ve already invested in making your app accessible, you’ve also already invested in making it testable.

UI testing also works great with UIs designed with size classes and localizations, including right-to-left support, which means your UI tests can scale to all presentations of your UI.

The new APIs yes, you can clap for that.

[ Applause ]

The new APIs and XE tests are available in Swift and Objective-C, and UI recording will generate whichever one you are using.

So these have been some of the many new features you will find in Xcode 7 as you build your apps for all of our platforms.

[ Applause ]

Now, we also have some exciting new technologies to share with you for games and graphics, and for that I’d like to invite my colleague up, Geoff Stahl.

[ Applause ]

GEOFF STAHL: Thank you, Matthew.

Xcode 7 looks amazing, and I am really excited to be here to talk about games and graphics.

It does not seem like so long ago that we were here introducing our first game-related API Game Center or social gaming network.

Since then, we’ve been busy with APIs for key input and graphics technologies, like game controllers, SceneKit, SpriteKit, most recently Metal.

This year, we are introducing a number of new game technologies, including things like Model I/O for modernizing your graphics; ReplayKit for extending your game’s social reach; and Game Center for adding super-smart brains to your games.

Also, we’re packaging all of these together into GameKit, so with one include, you get all this functionality at your fingertips, a great solution, great complete solution for game development across our platforms.

So let’s take a look at some of the new cool features in GameKit, starting with Metal.

So as you know, Metal is a 3D API we introduced last year for iOS.

We are really excited that this year we are bringing it to OS X.

And let’s, as a recap, look at an example we ran into just recently of game development moving from OpenGL to Metal.

So here you have a game when we started out on running on OpenGL, and what you notice is that OpenGL command processing is keeping one CPU completely saturated, which has pushed the game play to the second CPU, and the GPU is not very busy.

In fact, in this example, we saw the GPU was 40 percent idle.

Moving to Metal, we drastically reduce the command processing time.

We have an API that you can program more directly to the GPU.

And if you choose, you can split your command processing across multiple cores.

In this case, we’ve unlocked the full performance of the GPU.

We’ve been working with some third-party developers on Metal for OS X, and let’s take a minute to see what they’ve done on the Mac.

So it’s my pleasure to introduce David McGavran from Adobe Systems up to show what they’ve done with OS X in Metal.

David, welcome.

[ Applause ]

DAVID McGAVRAN: Thank you very much.

Thank you.

Adobe was very excited when Apple brought iOS to Metal’s iOS last spring.

In fact, we have already been using it to optimize some of our popular iOS apps like Premier Flip and Photoshop Mix and seen great results.

So we were thrilled when Apple told us they were bringing Metal to OS X.

This way we can share our code on both of those platforms.

So today I would like to show you some of our progress with two of our flagship Mac OS products, Illustrator and After Effects.

Let’s take a look.

So here we are in the current version of Adobe Illustrator with a massive piece of artwork with over 300,000 points, gradients, and blend modes.

I want to go ahead and do a zoom on this piece of artwork.

You are going to notice a bit of lag as it works its way into that complex piece of art.

We really wanted to see if we could do something better.

So with this version, we are demonstrating what we can do with Metal when we take the entire rendering engine and put it on Metal.

In fact, it was so performant, they were able to demonstrate a brand-new feature with continuous zoom.

So here you can see the results are amazingly different when you are looking at what we can do with Metal.

In fact, I can zoom all the way in here extremely far until you can actually read the text.

[ Applause ]

So this just drastically changed the way artists can work with our products.

Now we want to talk about After Effects.

So here is a video from extreme sports videographer Devinsupertramp.

Let’s take a quick look at this.

[ Music ]

That’s pretty fun.

What we want to do is take one of those shots into After Effects and add some effects into it to make a promo spot.

So here we are.

I am going to turn on some color correction, some ripples, and some lens flares, and go ahead and play that back without Metal.

And you can see while it’s working quite hard on the CPU, it’s not really giving that interactive performance we’d love to be able to give our compositors.

So we really challenged our programmers to take a look at this and see what they can do with Metal.

So in a very short time, they were able to port those three effects to Metal, and they are going to show you a pretty astounding difference.

So now I am going to switch to the same composition, but this time those effects have been replaced with the Metal versions, and I am going to play that back.

Immediately you are going to see it’s playing in real-time without dropping any frames, and we are seeing that in these effects up to an 8x performance improvement and drastically reduced CPU usage.

So Adobe is committed to bringing Metal to all of its Mac OS creative cloud applications, such as Illustrator and After Effects I showed you today, as well as Photoshop and Premier Pro.

We are very excited to see what Metal can do for our Cloud users.

Thank you very much.

[ Applause ]

GEOFF STAHL: David, thank you.

That is amazing.

I agree with Craig.

It’s really interesting to see when you can take an interaction that’s not real-time and move it to a silky smooth user interaction.

But Metal is not just for applications.

As we spoke about this morning, we are bringing key graphics technologies for both OS X and iOS onto the Metal API.

Up to now, technologies like core animation and our hardware accelerated Core Graphics, PDF and Safari page rendering, have been on top of OpenGL.

That meant the CPU was very busy feeding the GPU commands.

With Metal, we get improved efficiency and performance for critical user-level tasks.

Also, we didn’t want to just increase the system performance; we wanted to actually make it really easy for all of you to adopt Metal.

So we are introducing MetalKit.

So MetalKit is your gateway to Metal.

It does the heavy lifting for you.

MetalKit can do anything from set up your rendering loop to load 3D models via its interaction with Model I/O from your artist.

In fact, MetalKit can load meshes directly into Metal buffers and set up pretty much everything you can do to render your scene.

We didn’t stop there.

Metal performance shaders are high-performance image- processing shaders for Metal applications.

These are GPU-based parallel compute shaders individually tuned for all of our GPUs, again, making it really easy for all of you to adopt Metal.

So now you have a great rendering pipeline, but you need great content.

So we have Model I/O.

Sometimes loading and working with 3D models can be tricky.

So that’s what Model I/O comes in.

It makes it simple to load content from all the file formats you expect directly into SceneKit or Metal with this interaction with MetalKit.

But where it sets itself apart is the way it handles lighting.

Model I/O offers state-of-the-art, raytraced, materials-based lighting solution for you.

What this really means is it offers amazingly realistic graphics.

Let’s take a look.

So let’s start with a baseline.

This is just a model with textures loaded.

This is not really interesting.

It’s actually pretty flat, and I think the ’90s is calling this model back.

If we strip away the textures and apply a global illumination solution, what we see is something that immediately looks more realistic.

We add the textures back in, and we get something that looks really good.

Finally, of course, even with Model I/O and its lighting solution, you can add your own lighting in and shadows to get something that looks amazingly realistic, giving you those amazing results, from ordinary to extraordinary, Model I/O provides you a great solution for your models and for lighting.

So now you have a great rendering pipeline, you have great models, great content, you need great Gameplay.

So we have GameplayKit.

Gameplay is divided really into two areas.

We have navigation or how you move your objects in your game.

We also have your strategy or how your game thinks.

GameplayKit has solutions for both of these.

GameplayKit is a component-based API that is a full coverage of solutions for Gameplay.

Whether you need to track the gold from each of your players or need to navigate your bad guys from point A to point B and avoid the obstacles in between, or you need to generate deterministic random numbers for your next online multiplayer game, Gameplay has solutions for you.

From very simple games to very complex, we think Gameplay will do a very good job helping you guys develop your games.

We started out with a social piece, Game Center.

ReplayKit can extend the social interaction.

ReplayKit allows your players to record video replays of their gaming exploits.

They can save them or they can share them.

It’s hardware accelerated, it’s high performance, and it’s absolutely brilliantly easy for you guys to adopt, really easy to add it to your game, and maybe it’s the thing that kicks off that next viral marketing campaign for your game project.

Finally, SceneKit and SpriteKit.

We have some great updates for SceneKit and SpriteKit, but I think the most interesting, maybe the most compelling, is we have built tools for both SceneKit and SpriteKit directly into Xcode 7, and I’d like to hand it off to my colleague, Jacques Gasselin de Richebourg, to show you these tools inside of Xcode 7.

[ Applause ]


So last year with Xcode 6, we introduced visual scene editing for SpriteKit right inside Xcode.

This year we are kicking it up a notch and have added animation editing right inside Xcode.

So here we have a little cut scene I am making.

It’s got this cute red panda and he is about to make his way across the two chasms to the other side.

These are the base animations I have on there.

I want to layer these into the animation we expect.

First off, obviously, you need to make him run.

So let’s go down and grab an action.

Now, here we have a really interesting action.

It’s a reference action, which means it actually lives in a different file.

It lives in this actions.sks file.

I am referencing here so I can make composed actions.

So I am going to grab that.

Of course, it’s just as easy to just copy and paste.

I am going to do that a couple times, and naturally it appends after on the same track.

Let’s have a look at that, and we are going to use the scrubber this time, so it’s a timeline with a scrubber, and you can basically preview what the animation is going to look like at any time.

That’s not just a great preview tool; it’s actually great for editing as well.

So you noticed he didn’t jump over the chasms, and that’s a problem and something we are going to fix.

So let’s figure out a good spot for him to jump.

Looks like here is about perfect.

And then we are going to grab a move action, and we are going to build the jump right here.

Now, notice how it’s snapping to the timeline.

Okay. I am going to make it go up 100 units, and then I am going to use the timeline yet again to figure out the apex.

I want it to be about there.

Now I am going to snap the duration.

Okay? It’s that easy.

Obviously, when he jumps, I want him to ease out.

That means he is going to have a lot of velocity at the start and very little at the apex.

Going to copy and paste that and reverse it.

Minus 100.

That’s going to ease in so that he has some hang time up at the apex.

Okay. Let’s scrub across that.

That’s a parabolic jump.

Perfect. Now, obviously, it looks a bit weird if he just runs in the air, so let’s add an animation.

You notice here I am layering animations on top of each other, and SpriteKit handles that for you.

Here I am going to snap the animation to the length of the jump, and I am going to hop into the media library and fetch my jump frames.

So I am filtering jump, select that one, select all of them, drag and drop, perfect.

All right.

Now scrub.

Okay. That’s a cute jump.

Now, there are two chasms, so I am going to have to do this twice.

The easiest thing to do here is to turn this into a reference, so I am going to select all those, right-click, convert to reference, and I am going to call that Jump.

That’s going to store in the actions.sks file that I used the run from before.

All right.

Great. That’s a jump.

I’m going to use the scrubber again to lay down another jump.

The excellent thing about references is they turn up here in the object library straightaway.

So I am just going to drag and drop that in.

And here we should have our final cut scene.

Okay. Jump and jump.


[Applause] Okay.


The great things about references is they also help you structure not just your other scenes, but your code as well.

So here using SK action named, I am going to load up those very same actions, and I am going to hook them up to the left and right keys and the space bar for jump.

Let’s run that and have a look at it.

So here we have the panda doing the cut scene.


Now I am going to control it.

Look at that.

Left and right.

Jump, jump.

It’s that easy.

[ Applause ]

Now, we’ve been busy, so we took this to the next dimension as well.

So new in Xcode 7, we have full visual scene editing of 3D content using SceneKit as well.

This editor is very much like the 2D editor for SpriteKit.

Same things work.

Drag and drop to add new content.

You can use the manipulators in 3D to drag it around.

You can snap to other objects.

You can even duplicate.

And we have physics and animations live inside the editor.

So we are going to now animate.

Perfect. And let’s interact with this as well.

So I am going to play.

And we got the same cute panda from the previous demo.

Here we go.

[Music] This is a red panda, not a fox; a red panda.

I am going to walk up, and I am going to interact with that little block I had before.

Oh, yeah, kick you down.

Great. So in this game, you are playing a little red panda that collecting flowers and pearls and oh.

Sorry about that.

I promise, no pandas were hurt in the making.

Okay. This is not just a demo for stage here.

We are actually handing this out to you, and this is a sample that you can use to build your next 3D SceneKit game using this editor, SceneKit, and Xcode 7.

Thank you.

I am going to hand you back to Andreas.

[ Applause ]


Let’s quickly review the technologies we discussed in this session.

We announced new versions of our OSs today, and of course, the new native watchOS SDK.

The SDKs contain many new APIs you can take advantage of, particularly for gaming.

The new optimization techniques with App Thinning.

A new version of Swift that advances your apps faster and at the same time lets you write more effective code.

And of course, this new version of our Xcode now with user interface recording and testing.

And all these technologies will be available for download from the WWDC attendee portal this afternoon.

We hope that you will quickly adopt them all and create even more innovative apps for users.

Please go ahead and install all this stuff and let us know what you think about it.

Also we are doing something new with our TestFlight service for you this year.

Starting today, you will be able to deploy apps written to the iOS 9 SDK in TestFlight, and over the next several weeks, it’s even going to add support for App Slicing, on-demand resources, and Bitcode so you can test your apps even with those new technologies before we launch iOS 9 later this year.

[ Applause ]

So there are, of course, many opportunities to learn more about these technologies here at the conference.

There’s more than a hundred sessions you can attend, numerous labs where you can go and get help with your projects.

In fact, there’s more than a thousand Apple engineers here on site to answer your questions.

So I hope you enjoyed this session.

I hope to hear from you later this week.

[ Applause ]

Apple, Inc. AAPL
1 Infinite Loop Cupertino CA 95014 US