Best Practice for Sending Windows 8 Tiles from Mobile Services

Those that know me know I am not a fan of javascript, in pretty much all of its forms (including node.js), however, I’m really digging Windows Azure Mobile Services (WAMS).  WAMS allows you to easily provide a back end to applications for storing data, authenticating users, and supporting notifications on not just Windows and Windows Phone, but also iOS with future plans of supporting Android soon.

Now, I mention javascript because WAMS provides a slick node-like powered data service that makes it really easy to store data in the cloud.  The ToDoList example exercise illustrates the ease at storing user data in the cloud and hooking it up with authentication and notification support.   The nice thing about the authentication is that it’s easily integrated into the backend:

image

But, more on this later.  Right now, I want to deal with notifications in WAMS.  In WAMS, you have the opportunity to right custom server-side javascript to do things like send notifications on insert/update/delete/read access:

image

In my case, I want to send a tile update if the new data meets some criteria.   Let’s start all the way down the code and work our way out, starting with the notification piece.    One page you MUST have bookmarked is the tile template catalog on MSDN.   This page defines the XML syntax for all possible tiles your tile can have, including both small/square tiles, and large/wide tiles.    All of these have a defined schema, such as this for TileSquarePeekImageAndText04:

 <tile>
  <visual>
    <binding template="TileSquarePeekImageAndText04">
      <image id="1" src="image1" alt="alt text"/>
      <text id="1">Text Field 1</text>
    </binding>  
  </visual>
</tile>

Which produces a tile that “peeks”, such as this (which flips between the top half and bottom half):

TileSquarePeekImageAndText04 example

Yes, it’s easy to laugh at the magic “04” in the template title.  I like to joke that my personal favorite is TileWideSmallImageAndText03.   But, there variety is crucial to creating the ideal app experience and that depends on how you want to display the data -- and that requires knowing the XML template.

Now, in WAMS, there’s a great tutorial on sending some basic notifications.   In that walkthrough, a notification is sent via the server-side javascript like so:

push.wns.sendTileSquareText02(“https://bn1.notify.windows.com?[snip]”, { text1: “some text”, text2: “more text”});

Now, at first glance, this is very nice because WAMS will write the XML for you.  However, you still must know what data the template requires.  Does it need an image?  One text line?  Two?   You get the point.  Unsurprisingly, calling that method will generate XML like:

 <tile>
  <visual>
    <binding template="TileSquareText02">
      <text id="1">some text</text>
      <text id="2">more text</text>
    </binding>  
  </visual>
</tile>

You can learn more about this in the WAMS script reference.  Another must-have bookmark.  However, I recommend you don’t use these at all, and instead write the XML payload directly.   This is for a few reasons, but primarily, it’s for control – and, really, you have to know the fields required anyway and you’ll still have the tile catalog page open for reference.

In looking at the mpns (Microsoft Push Notification Service) library a bit closer (awesome job by the guys, by the way) up on git, it has this method:

 var raw = new mpns.rawNotification('My Raw Payload', options);

When developing my app, I realized I had no idea what tile size the user has.   Some may opt to use a wide tile, others a small tile.  I needed different tiles to support both.   I didn’t like sending two notifications (seems wasteful, doesn’t it?) and to do this efficiently, it’s easier to just create the payload explicitly that includes all tiles.  For example, this includes two completely different tiles:

 var payload = "<tile><visual><binding template='TileWideImageAndText02'>" +
      "<image id='1' src='" + xmlEscape(bigImage) + "' alt='map'/>" +
      "<text id='1'>" + text1 + "</text>" +
      "<text id='2'>" + text2 + "</text>" +
      "</binding>" +
      "<binding template='TileSquareImage' branding='none'>" +
      "<image id='1' src='" + xmlEscape(smallImage) + 
      "' alt='map'/></binding></visual></tile>";

push.wns.send(channelUri, payload, 'wns/tile', 
{
        client_id: 'ms-app://<snip>',
        client_secret: 'i will never tell',
        headers: { 'X-WNS-Tag' : 'MainTile' }   
}...

Sure, it doesn’t look as clean and (gasp!) we have to do string concatenation.   But, it’s only a couple of minutes more work and just more flexible.  Like I said: either way, you need to know the template.   In my case, I’m sending both notifications in one payload.  The first is TileWideImageAndText02, which produces a nice image with the text on the bottom describing the image.  If the user has a small tile, it will use TileSquareImage, which basically just forgoes the text and just displays the image.  After trying a few, I settled on this combination as the best user experience.  This is an easy way, with minimal effort, to support both wide and narrow tiles.

As an aside, I recommend setting the tag (X-WNS-Tag) header, particularly if your app cycles tiles and you want to replace a specific tile.  Also, it’s a good ideal to XML escape all data, which I’m doing with the long image URLs … and this, I believe, is taken right from the mpns library:

 var xmlEscape = function (text) {
    return text.replace(/&/g, '&amp;')
       .replace(/</g, '&lt;')
       .replace(/>/g, '&gt;')
       .replace(/"/g, '&quot;');    
}

If you don’t escape the data and have some illegal chars in there as a result, the notification gets sent correctly (that is, accepted), but gets ignored by the client.

Now that I’ve got the basic code to send a tile, I needed to filter some data and run a query and sort users by distance.  Sound like fun?  I’ll write about that next…