Working with data in Flex

This session was led by James Ward. I was somewhat familiar with him having found a couple of interesting posts on his blog in the past about easing functions in Flex. He focused mainly on what you should use to store and move your data under different circumstances. He briefly discussed the Tour de Flex AIR application which everyone in the session was given a copy of on a nice 1gb thumb drive. Tour de Flex shows how to use a myriad of components, both built in Flex components and third party ones. I was of course interested in playing with the mapping APIs included from Google, MapQuest and Yahoo Maps.

Next we talked about different ways to send your data over the wire from the server to the client. James has a great Flex application available online which allows you to test out all of the different options. Of course AMF (Remote Objects) was basically the fastest. It did seem to get beaten by Data Services using Flex Paging though, but that's what you pay for with Blaze DS I suppose. The application is quite nice and allows you to send rows across from a server side database.

He followed that presentation by giving a few other examples that boil down to things to be careful of when dealing with large amounts of data. With XML there is a memory overhead which can become significant with large amounts of data and can cause performance issues.

For AMF (surely nothing is bad about AMF! It's a binary, no parsing required, way to send data!) he said to watch out for circular references and the overhead you accept for the use of binding. Sounds like a no big deal thing to me but maybe Adobe pays him to get people to consider buying Blaze DS?

One last thing he talked about was something I had assumed but never really quantified. He built a quick Flex application which loaded 100,000 objects into an Array and the same objects into an ArrayCollection. It took the Array a mere 100 milliseconds to complete where the ArrayCollection took 1.4 seconds! So if you are dealing with large amounts of data and want your application to be fast, watch out for those ArrayCollections.