Blue Flex

The application that I’m working on with the guys at Jampot has made the BBC news. Check out the article from the BBC.
http://www.bbc.co.uk/news/uk-northern-ireland-15075450

Also the guys are heading over to the Adobe Max 2011 conference so you’ll be able to see it live and see how easy it is to create a mobile app for all mobile platforms.

Developing for Apple’s devices can throw up a few little quirks that don’t happen when using Android devices.

This one happens if you are using shared objects to store information between sessions.
Basically, you should always call the flush mechanism whether you are adding more data to the shared object or if you are deleting something from the shared object.

What you find is if you have a shared object ‘shared’ with a value shared.data.firstValue = “something”, then you delete that value using

delete shared.data.firstValue;

if you try to access the value firstValue you will get null.
This is exactly what I’d expect.

Then lets say you exit the app and you either kill the app from running in the background or iOS stops it. Then the next time you load the app and access the shared object shared.data.firstValue you will get back “something” and not null.

You must flush the shared object for it to be stored locally, otherwise when the app is killed, the local storage will not have been updated.

Its quite a common thing with Flex and actionscript projects to create an Object and inside that object it will have many properties.  Something in your view will be bound to the object so that the view changes with the object. So long as you change the entire object this will work fine.

Where this doesn’t work is if you change a property inside the object.

So if we have something like this

[Bindable]			
private var myObject : ObjectDataVO;
 
<view:SomeComponent
...
data="{ myObject }"
...
/>

When we set myObject to something the view component gets updated (great so far).
Lets say the myObject has a property text and the view component uses this to display some visual label, then somewhere in the app I change that property, myObject.text = “something else”;
The binding will not trigger as I haven’t actually changed the myObject, just a property inside it.

So how do we fire the binding manually? Well there is the BindingManager class (note this is an excluded class so you’ll not see it in the autocomplete ).
So in this example if I changed the myObject.text property then I could call

BindingManager.executeBindings( this, ‘myObject.text’, myObject );

This would fire of the binding as if the actual myObject had changed so anything listening in will now get updated.

Sometimes I like to put something up on my blog that’s more as of a bookmark for myself as I know I’ll want to look it up at some point.  So what I’ve got here is a slack, open cross domain policy.

DO NOT USE THIS IN YOUR PRODUCTION CODE
(unless you really need to and understand why you shouldn’t)

This will get rid of any security issues you may be having while in development.

<?xml version="1.0"?>
<!DOCTYPE cross-domain-policy SYSTEM
 "http://www.adobe.com/xml/dtds/cross-domain-policy.dtd">
<cross-domain-policy>
<site-control permitted-cross-domain-policies="all"/>
<allow-access-from domain="*" secure="false"/>
<allow-http-request-headers-from domain="*"
 headers="*" secure="false"/>
</cross-domain-policy>

Working on a mobile project I needed to create a renderer for a list, so I choose to look at the IconItemRenderer which extends the LabelItemRenderer. These have been optimised for mobile use so it seemed a reasonable place to start. On the whole they seem like good classes to use, but if you’ve ever worked with the Datagrid/DataGridBase in the past you will probably know about the white square which comes about from the hardcoded #FFFFFF values inside the DataGridBase!

Well the IconItemRenderer and LabelItemRenderer have a similar issue. So lets just say you create a list and you wish to skin the list exactly how you like or use it in a tile layout or something other than vertical then you will find some lines above and below your renderers which look out of place. You can’t get rid of them no matter what property styles you set.

The fix is pretty straight forward but why does there have to be some hardcoded values in something that is meant to be very versatile?

So inside the LabelItemRenderer around lines 881 you will see the following. It uses these values to draw separators whether you like it or not.

// separators are a highlight on the top and shadow on the bottom
topSeparatorColor = 0xFFFFFF;
topSeparatorAlpha = .3;
bottomSeparatorColor = 0x000000;
bottomSeparatorAlpha = .3;

So the quickest way of dealing with this is to override the drawBackground function in your own class which is created in LabelItemRenderer. This doesn’t get called from IconItemRenderer so you can quite simple copy the entire function and just remove the separator chunk and do not call super from your function which overrides the drawBackground.

Better still would be to change the hard coded values to styles from a CSS file.

var topSeparatorColor : uint = getStyle( 'topSeparatorColor' );
var topSeparatorAlpha : Number = getStyle( 'topSeparatorAlpha' );
topSeparatorAlpha = isNaN( topSeparatorAlpha ) ? 1 : topSeparatorAlpha;

If you’re setting a Number just remember to check for NaN’s in case you haven’t set a style, uints default to 0 anyway.

The other day I wanted to create a function in a class that would take a Function as a parameter and an Array of arguments.  Much like callLater() does, but not doing the whole queuing thing until the next frame.

So how do you call a function that may have any number of arguments. Well here is the code and it should speak for itself.

 protected function checkSomethingThenCallOtherFunction(
    method : Function, args : Array = null ) : void
 {
    if( something.length < someMaxLimit )
    {
        method.apply( null, args );
    }
    else
    {
        //do something else
    }
}

So how easy is that? Pass in the function and an array of arguments, that’s all.

I was playing around with some code recently (Flex 4 code) and I went to create a simple background and not having Catalyst or similar to output a fxg file I went to create my own gradients with some code. After a couple of goes and not getting anything resembeling what I expect I decided to write a quick explorer.

I’ve done something similar ages ago with flex 3, so I thought I’d do this with flex 4 and perhaps look to expand it as an example of reskining an app with different skins. (source code may follow when I do this)  So here is the first step. A simple explorer to help understand the values that are used to make a Linear or Radial gradient along with the entries that make the look how they look.

I think it should be self explanatory, but if not just post a comment.

Follow the link to open the explorer.

Many times you look something up, do it once and think cool I’ll remember that as it’s simple. Then 1 year later you’ve forgotten the syntax and you can’t find that help/blog page where you learned about it the first time.
Well I needed to add in some debug code that would only be there for debugging, and the last thing I want to do when building a release version is to scan through the code to remove it. So the ideal way is to use a conditional compiler argument.

So in Flashbuilder, under the project properties and then the Flex compiler properties you’ll see something like this

Example for custom arguments

So if you had the following defined, -define=CONFIG::DEBUG,true -define+=CONFIG::SOMETHING_ELSE,false

Then in code you could do the following.

    CONFIG::DEBUG
    private var test : Boolean;
 
    CONFIG::SOMETHING_ELSE
    private function somethingElse() : void
    {
 
    }

The variable and function code will only be included if the compiler argument is true. So in the above example if you called the function ‘somethingElse()’ then this would generate a build error as somethingElse() doesn’t exist. Change the argument to true and it will build fine.