jdeveloper

Adding Apple Watch to a MAF App

Posted on Updated on

Here are some steps for adding iOS 9+ Apple Watch OS 2+ support to an Oracle Mobile Application Framework (MAF) App.

Set up and deploy the MAF app

1. Download and unzip the following Cordova plugin zip from git: https://github.com/leecrossley/cordova-plugin-apple-watch/archive/master.zip (more info at https://github.com/leecrossley/cordova-plugin-apple-watch)

2. Using JDeveloper, create a MAF app and select the unzipped plugin folder in the maf-application.xml (in the “Additional Plugins” section)

3. Create an AMX page (using amx:clientListener in amx:commandButtons to trigger the plugin APIs); this can be just placeholder content for now

4. Deploy the app to the simulator to generate an Xcode application in the JDev app’s “deploy” folder. We will be editing this generated output from the deploy folder so be careful to note that generated output will be lost if you re-deploy from JDev in the future.

5. Quit JDev

Prepare Xcode

6. Open the Xcode project from the MAF app’s deploy folder (Oracle_ADFmc_Container_Template.xcodeproj).

7. Create a new Target – WatchKit App (ensure “Notification Scene”, “Glance Scene”, and “Complication” are checked and activate the new scheme when prompted)

8. Open the Interface.storyboard from the newly-added target

Create the Watch App’s UI

9. Drag a Label from the Object Library onto the Interface Controller Scene’s Interface Controller and use the Attributes Inspector to style the label if desired (e.g. Size – Width and Height = Relative to Container and value 0.7)

Drag a Button from the Object Library onto the same scene below that Label.

10. From the watch extension folder in Xcode, open the InterfaceController.swift file in a new window or use the “ven diagram” (assistant editor) icon for split-screen editing.

11. While holding down the Control key, drag the label from the scene hierarchy that was added in step 9 into the class of step 10 to generate an Outlet variable:

@IBOutlet weak var helloWorldLabel: WKInterfaceLabel!

Repeat with the button to generate:

@IBOutlet weak var someButton: WKInterfaceButton!

and then also type in:

@IBAction func someButtonAction()
{
  print("someButtonAction")
}

Finally, Control-drag the hollow circle from the someButtonAction margin onto the button in the scene to connect the action to that button.

12. If you want to create a page-based app, drag additional Interface Controller scenes from the Object Library and connect them via Control-dragging the seque line from the selected scene to the new scene (and then in the seque panel, choose “next page” as the relationship)

Make live data to test

13. In the InterfaceController.swift file, add this temporary code to the end of the awakeWithContext method so we can verify the app functions before connecting it to the MAF app via the Cordova plugin:

// Create a timer to refresh the time every second
let timer = NSTimer.scheduledTimerWithTimeInterval(1.0, target: self, selector: #selector(InterfaceController.updateLabel), userInfo: nil, repeats: true)
 timer.fire()

14. Add this additional function to the class to support the temproary timer code:

func updateLabel()
{
  print("updateLabel")
  let time = NSDate()
  let formatter = NSDateFormatter()
  let timeZone = NSTimeZone(name: "UTC")
  formatter.timeZone = timeZone
  formatter.dateFormat = "h:mm:ss a"
  var formattedString = formatter.stringFromDate(time)
  formattedString = formattedString.stringByReplacingOccurrencesOfString(" p", withString: "PM", options: [], range: nil)
  formattedString = formattedString.stringByReplacingOccurrencesOfString(" a", withString: "AM", options: [], range: nil)
  self.helloWorldLabel?.setText("UTC \(formattedString)")
}

15. Run the WatchKit target to verify everything was created properly.

16. Stop the app in Xcode

Configure the Cordova Plugin APIs and, in turn, MMWormhole

17. Go to Oracle_ADFmc_Container_Template – all 3 targets: the phone, the Watch Target, and the Watch Extension Target – Capabilities – App Groups and turn it on. Ensure that all 3 checkmarks are checked. Your Xcode identity and Bundle Identifier must allow you to use app groups. This is a requirement of the MMWormhole framework (you’ll get an error and crash when running the app if you don’t do this). The “App Groups” must have an entry with a value that you will later use in your Swift code (step 23) and JavaScript code (step 25).

18. There is no need to manually add MMWormhole (https://github.com/mutualmobile/MMWormhole) to the Xcode project because the Cordova extension already has it included. However, you will need to select all of the MMWormhole*.m files in Xcode from the Plugins folder and then add (checkbox) target membership for the watch extension.

19. By default, the watch extension will not be able to see those MMWormhole Objective-C objects so you need to create a bridging header. Go to Oracle_ADFmc_Container_Template – Choose the Watch Target Extension – Build Settings – Swift Compiler Code Generation – Objective-C Bridging Header and specify the value:

Oracle_ADFmc_Container_Template/Plugins/Bridging-Header.h

20. In the Navigator, select the Plugins folder and choose File – New – File – Header File and name it Bridging-Header.h and choose the Watch Target Extension target and then use this as the file content:

#import "MMWormhole.h"
#import "MMWormholeSession.h"

21. In the watch extension InterfaceController.swift file, comment out the existing 2 lines for the timer in the awakeWithContext method (from step 13).

22. Add the following to the awakeWithContext method of the watch extension InterfaceController.swift file:

print("awakeWithContext")
let watchConnectivityListeningWormhole = MMWormholeSession.sharedListeningSession()
watchConnectivityListeningWormhole.activateSessionListening()
watchConnectivityListeningWormhole.listenForMessageWithIdentifier("from_phone_queue", listener:
{
  (messageObject) -> Void in
    if let message: AnyObject = messageObject
    {
      print("from_phone_queue")
      // handle your message here
      let messageValue: String! = message.objectForKey("selectionString") as! String
      self.helloWorldLabel?.setText("Phone \(messageValue)")
    }
})

23. Add this to the someButtonAction method of the watch extension InterfaceController.swift file:

// Make sure this appGroupId value is
// specified in the App Groups config from
// step #17
let appGroupId = "group.com.yourfoo.yourbar"
let wormhole = MMWormhole(applicationGroupIdentifier: appGroupId, optionalDirectory: nil, transitingType: .SessionContext)
let diceRoll = Int(arc4random_uniform(6) + 1) // 1 through 6
wormhole.passMessageObject("rolled \(diceRoll)", identifier: "from_watch_queue")

24. Clean then build to make sure the MMWormhole objects compile, otherwise the above steps need to be revisited.

25. Edit the app’s AMX page so that it has these components to interact with the Cordova plugin APIs:

<amx:verbatim id="v1">
<![CDATA[
window.handleMessageFromWatch = function(msg)
{
  console.log("handleMessageFromWatch");
  var outputTextElement =
    document.getElementById("ot2");
  if (!outputTextElement)
    console.log("Unable to find ot2");
  outputTextElement.textContent =
    "Watch " + msg;
};

window.handleAWInit = function(clientEvent)
{
  console.log("handleAWInit");
  var appGroupId = "group.com.yourfoo.yourbar";
  var queueName = "from_watch_queue";
  applewatch.init(
  function successHandler(appGroupId)
  {
    console.log("init Success", appGroupId);
  },
  function errorHandler(foo)
  {
    console.log("init Failure", foo);
  },
  appGroupId);

  applewatch.addListener(
    queueName,
    window.handleMessageFromWatch);
};

window.handleAWSendMessage = function(clientEvent)
{
  console.log("handleAWSendMessage");
  // The value of queueName must match a value
  // used in a corresponding listener in Swift
  // or Objective-C within the watch app using
  // the method
  // MMWormhole.listenForMessageWithIdentifier
  // in order for the watch to receive the
  // sent message.
  var diceRoll = // 1 through 6
    Math.floor(Math.random() * (7 - 1)) + 1;
  var message = "rolled " + diceRoll;
  var queueName = "from_phone_queue";
  applewatch.sendMessage(
    message,
    queueName,
    function successHandler(foo)
    {
      console.log("send Success", foo);
    },
    function errorHandler(foo)
    {
      console.log("send Failure", foo);
    });
};
]]>
</amx:verbatim>
<amx:commandButton id="cb1" text="Init">
  <amx:clientListener id="cl1" type="action"
    method="window.handleAWInit"/>
</amx:commandButton>
<amx:commandButton id="cb2" text="Send Message">
  <amx:clientListener id="cl2" type="action"
    method="window.handleAWSendMessage"/>
</amx:commandButton>

25. Run the app.

If you get an error about “The operation couldn’t be completed. (LaunchServicesError error 0.)” then look for the hidden error details in this file:

~/Library/Logs/CoreSimulator/CoreSimulator.log

In the phone app, click “Init” then “Send Message” to see a message appear in the label in the watch app. In the watch app, click the button to see a message appear in the phone app.

Final Notes

Remember that you are working in the deploy folder of JDeveloper so the next time you deploy, you’ll lose all your edits to the Xcode project. While this process involves editing files of the “deploy” folder of the MAF app, you’ll need to take careful notes on what changes you make in the deployed folder in case you need to re-deploy in the future (and you will need to) since that would require you to perform these steps again. Perhaps there is some way to bundle all the WatchKit content into a Cordova plugin so you could completely avoid having to mess around with the deploy folder (I have not researched this but if you have an have a blog post to share with detailed instructions, please let me know so I can add a link to it in this post).

A real app wouldn’t expose an “init” button to the user so you’d want to move that init call into a “showpagecomplete” handler like the one seen in the MAF developer guide. Typically a JavaScript feature include file would be where this code would go (just like the custom component code seen in the CompGallery public sample app).

Mobile Web Inspector Developer Tools

Posted on Updated on

What is a web inspector?

A web inspector is a developer tool that allows you to debug JavaScript, CSS, and HTML. You’ll usually get a file resource browser, performance monitoring tools, the ability to set code breakpoints (although in my experience breakpoints are somewhat flaky depending on which version of the mobile OS and desktop browser you are using–so often I resort to using the JavaScript “debugger” keyword to act like a breakpoint), the ability to edit HTML, and a tool for visualizing, resolving, and changing style properties–all without having to redeploy your application.

Apple introduced the ability to use the desktop Safari web inspector developer tools with iOS application WebViews and mobile Safari back in iOS 6 and Safari 6. Each new release of iOS has required that you use a corresponding version of desktop Safari so beware of this if you are planning to upgrade one device but not the other.

Google introduced the ability to use the desktop Chrome web inspector developer tools with Android application WebViews and mobile Chrome in Android 4.4 when the Chrome WebView was introduced. Many legacy Android devices are still being sold (let alone in use) but unfortunately, it is not compatible with the web inspector. For situations like that, there are some third party techniques available that require you to inject code into your application in order to send messages back and forth with a server. Those solutions have much fewer abilities than a formal web inspector. To combat this legacy issue, Google introduced a Google Play-updatable WebView in Android 5 Lollipop so your application can take advantage of new technologies even if the OS is not upgradeable.

Regardless of the platform, it is important to note that the WebView and the mobile browser are not equal. If something works in the browser, you cannot always expect it to work exactly the same way in the WebView. For example, on Android, I have seen as much as a major version number difference of 3 between the two. On iOS (prior to WKWebView), JavaScript performance was severely throttled compared to what mobile Safari had.

This post only discusses web inspectors. If you are interested in debugging Java code for your application, refer to the Oracle MAF developer guide for instructions.

Setting up your Oracle MAF Application

For iOS, there is no property to edit, you either need to use a debug deployment or use Xcode to run the application. For Android, you must edit the src/META-INF/maf.properties file of your application so that it defines:

javascript.debug.enabled=true

You can only use a web inspector with development (aka non-production) applications.

If you want to connect the web inspector before the first screen loads, will need to edit your run configuration’s “Application Arguments” to specify the following which allows you to configure how many seconds before the WebView will get populated with content.

-featureContentDelay 20

If doing this in an already-generated Xcode project, this setting goes in the Run Scheme’s “Arguments Pass On Launch” field.

Setting up your Desktop Browser

I assume that you already have your device set up for development so I won’t cover it here (e.g. Android needs development mode and USB debugging turned on via tapping the build number 7 times, iOS needs the device set up as “Use for development”, etc.).

Desktop Chrome doesn’t require anything special but I use this optional “ADB” Chrome extension as a convenience since I find it easier to click an icon vs. typing a special URL in the address bar.

Desktop Safari requires you to enable the “Develop” menu in its Advanced Preferences dialog:

Safari Preferences Advanced Settings

Using the Safari Web Inspector

From the Safari Develop menu, your connected devices and iOS Simulator will display as submenus. Each submenu will list all of the WebViews that you are allowed to inspect (if you are not sure which one you want, a light blue highlight will appear on your device or in the simulator as you hover over each menu item).

Open an iOS WebView inspector

A web inspector window will open once you select that menu item:

Safari Web Inspector

The UI varies across versions of desktop Safari but I’ll describe one here as an example. The “Resources” tab (1) is where you can browse the DOM, look at scripts, style sheets, etc. When the HTML resource is selected, the main panel will show you the DOM tree (3) which you can manually browse for an element. Alternatively, you can use the crosshairs tool (2) to click on the screen (of the actual device or the simulator) to pick an element that you want to inspect.

With an element selected, you can then use the “Styles” tab (4) of the right panel to view the element’s style properties. This will show you the exact selectors that are providing each style to the inspected element (5). This is incredibly valuable, particularly when multiple selectors are competing (a line will strike out any styles that lose out due to some other more specific selector). You can also edit those values directly or add additional properties and see the change immediately take effect on the screen. Beware changes made in the inspector do not get saved permanently into your application. If you are trying to create a custom style sheet, you’ll need to jot down elsewhere the changes you’ve made; if you leave the page or close the app, you’ll lose your changes.

Using the Chrome Web Inspector

First, visit Chrome’s “chrome://inspect/#devices” page to see a list of your connected and debuggable Android devices. Each will appear with all of the WebViews that you are able to inspect. An icon will appear with a darker gray and the WebView listed towards the top will be what is currently visible on the Android screen.

Chrome Devices

A web inspector will open once you click the “inspect” link.

Chrome Web Inspector

The UI varies across versions of desktop Chrome but I’ll describe one here as an example. The “Elements” tab (1) is where you can browse the DOM. The main panel will show you the DOM tree (3) which you can manually browse for an element. Alternatively, you can use the magnifying glass tool (2) to click on the screen (of the actual device or the emulator) to pick an element that you want to inspect.

With an element selected, you can then use the “Styles” tab (4) of the right panel to view the element’s style properties. This will show you the exact selectors that are providing each style to the inspected element (5). This is incredibly valuable, particularly when multiple selectors are competing (a line will strike out any styles that lose due to some other more specific selector). You can also edit those values directly or add additional properties and see the change immediately take effect on the screen. Beware changes made in the inspector do not get saved permanently into your application. If you are trying to create a custom style sheet, you’ll need to jot down elsewhere the changes you’ve made; if you leave the page or close the app, you’ll lose your changes.

Further Reading

https://github.com/phonegap/phonegap/wiki/Debugging-in-PhoneGap

Accessing Managed Beans from Java

Posted on

The following link shows a helpful article comparing alternate ways of using Java to programmatically retrieve an ADF Faces or JSF managed bean.  In other words, this is how to evaluate EL in Java code.

http://java.sys-con.com/node/1984284

iOS Web Inspector

Posted on Updated on

The content of this post has been updated with additional detail at:

https://formattc.wordpress.com/2015/08/21/mobile-web-inspector-developer-tools/

Creating a panelGridLayout or tableLayout structure

Posted on Updated on

It may be tempting to go crazy with panelGroupLayout components in ADF Faces and ADF Mobile. The panelGroupLayout is not the universal hammer. In many cases, you will have a simpler structure (easier to maintain, more predictable, faster performing, more flexible) if you build your structure using a panelGridLayout or tableLayout instead.

structure

When you are given a mock-up of what the application should look like, you may not know how to break it down into something simple. Either print out the design on a piece of paper or open it up in a graphics program where you can draw some lines.

  1. Draw vertical lines representing potential column divisions (red)
  2. Draw horizontal lines for potential row divisions (green)
  3. Now that you have a basic grid structure, use a different color (yellow) to draw x marks where you see cells that need to span multiple columns or rows:
    Initial set of lines
  4. You may find that your column lines really don’t make sense (in this case, the 2 middle columns had no cells in them that didn’t need to span into a nearby column. This is an indication that there probably ought to be 2 separate grids or tables.  Use a new color (magenta) to draw a line where the division makes sense and repeat the process again:
    Refined lines

Now you can go back to coding your gridRow/gridCell or rowLayout/cellFormat components because you can visually see where the content goes and where you need to use a special columnSpan or rowSpan. As an added bonus, you now have the ability to specify the sizes for your cells as well as the horizontal and vertical alignments of your cells–something panelGroupLayout would not allow you to control.

Final structure

Notice that the labels and fields of grid 2 (columns 1 and 2) are showing separate components. This is done so that there can be matching horizontal alignment among the fields of the two rows. If these were input or select components, you would need to use simple=”true” (to hide the built-in label) and then use separate outputLabel components that reference them.

While we are only using consecutive grids in the example discussed in this article. You might come across a need to nest a grid inside of another grid. Do not excessively nest panelGridLayout nor tableLayout components (if you have 3 or more deep, consider a re-evaluation of your structure). You may find it helpful if someone else reviews your structure. Your goal should be to create something simple.

Mac JDeveloper – Show in Finder

Posted on

Ever wanted to easily reveal a file that you’re working on in JDeveloper in the Mac OS X Finder?

Here’s how to set up a toolbar button and context menu item to do just that:

  1. Go to Tools – External Tools
  2. Click New…
  3. Choose these settings:
    Type = External Program
    Executable = open
    Arguments = -R ${file.path}
    Caption = Show in Finder
    Add Item to Menus = Tools Menu, Window Context Menu, Source Editor Context Menu
    Add Buttons to Toolbars = Main Toolbar
    Log Output to Messages Log = unchecked
    Enabled = When a file is selected or open in the editor

Once you’ve created that tool, simply right-click a file in the application navigator, right-click an editor, or use the toolbar button to switch to have JDeveloper automatically switch over to the Finder and open a window with that file selected.

HiDPI Images in MAF or ADF Mobile Applications

Posted on Updated on

Whether you are developing a MAF (or ADF Mobile) application or a classic ADF Faces application, it is becoming more and more important to support HiDPI screens (high resolution or “retina” displays). There is no easier way to make your application look out-dated than to use grainy, unprofessional image assets or use them improperly.

In HTML, the “px” unit (sometimes referred to as a “dp” or density-independent pixel) corresponds to the same amount of space regardless of the DPI of your screen (however, the number of screen pixels may vary). The original iPhone models did not have HiDPI displays. Each point of color on those screens corresponds to one HTML CSS “px” unit. Newer iPhone models introduced a HiDPI (or “retina”) display that has 4 screen pixels in the same amount of space that 1 screen pixel used to take up (2 screen pixels wide by 2 screen pixels tall); on these new screens, the width and height “px” values use twice the amount of screen pixels.

Why is this a common problem? Since MAF (or ADF Mobile) uses HTML, displaying an image is not as simple as just specifying the source path and magically hoping the browser will know that you are using a high resolution image. You must specify dimensions to go along with the source path.

Let’s work with an example. You have an image named “image-64.png”. This image has a size of 64 by 64 individual dots of color (individual points of color information). If you coded your page like the following, the image will be shown with a width of 64px and a height of 64px (one color dot per “px”):

<amx:image id="i1" source="images/image-64.png"/>

This would look just fine on a classic low-DPI display. However, on a HiDPI display, it still takes up the same space but since there are more screen pixels, the image will look very grainy.

In order to look crisp and professional, you need to set a size so that each dot of color corresponds to at least one screen pixel. For a HiDPI display, this means your image needs a width and a height specified such that you use 2 dots of image color information per HTML CSS “px” unit (e.g. a 64 by 64 sized image should be specified to use a width of 32px and a height of 32px. In code, your page should look like this:

<amx:image id="i1" inlineStyle="width:32px;height:32px" source="images/image-64.png"/>

Even if you still want to support legacy screens for your application, this same image (with the same specified width and height) will look beautiful on low-DPI screens because of how images are processed modern browsers.

If for some reason you really needed or wanted to specify alternate images for each kind of screen, you have the option to use a screen properties EL variable to toggle the rendered state of alternate amx:image components or simply use that EL to alter the inlineStyle and the source path as desired.

PNG vs. SVG?

Consider whether using PNG is really the right choice to begin with. SVG (scalable vector graphics) files tend to be smaller file sizes when created without excessive anchor points, no embedded bitmap data, reduced decimal number precision, excess whitespace removed, and comments removed. If the resolution of your display gets better, the SVG will look that much better; you don’t need multiple resolution versions of the same image resources. If the SVG was given the proper width/height to begin with by its creator, you won’t even need to muck with inlineStyles since the image will naturally be displayed at the proper size.