I attended my first meeting of the Android Australia User Group - Sydney this evening.
The organiser, John Scott, started with some general announcements, which should be of interest to many.
- There is an Amazon Cloud Drive Developer Tech Talk in Sydney on Monday October 10, 2011.
- Google Developer Day will be on November 8, 2011 in Sydney.
Those interested should register for these events. They're both free, as far as I know.
John then kicked off the first talk of the evening, which was a brief overview of Motorola's IDE for Android application development - MOTODEV Studio. According to John, this IDE is better than the Google Android plugin for Eclipse, which seems to be the de facto IDE for Android, because it has a few extra features like the generation of boilerplate code to make the developer's life easier, and also graphical tools to manage SQLite databases. I found a couple of other discussions on MOTODEV Studio vs Eclipse here and here, but they seem somewhat dated.
The second talk was by Gianpaolo De Biase, a developer with AppCast. Gianpaolo talked about some real-life applications that his company has built (one of which is JustStartWalking, an app for an initiative of the same name from the Chiropractors' Association of Australia). He also discussed two important architectural decisions that his company made when developing these products.
1. When dealing with local databases, Gianpaolo recommends using an Object-Relational Mapping (ORM) tool rather than the raw SQLite API provided by Google. The tool he used is OrmLite, which is Open Source like Hibernate but more lightweight and adequate for the simpler data structures needed for local storage on mobile devices.
2. When making REST calls to remote servers, he recommends using the Spring Android module with its REST Template in preference to raw HttpClient. He did encounter some bugs with Spring Android in the areas of cookie management and SSL certificates, but believes the product is rapidly maturing.
As a side benefit of both OrmLite and Spring Android being annotation-based, Gianpaolo was able to have a single set of domain objects. Each object had two sets of annotations applied, one for persistence and one for interfacing with REST services. [I'm a bit suspicious of the latter annotation, since it looks like the design combines domain objects and message documents into a common entity, a form of tight coupling I've long warned against.]
After a short break, we had our third talk of the evening by James Zaki, a freelance developer with goCatch. This was a high-level description of the goCatch app, which brings together cab drivers and prospective passengers. There was general satisfaction expressed around the room in favour of this cartel-breaking app, since the taxi companies and CabCharge engage in significant rent-seeking behaviour at the expense of both drivers and passengers. goCatch allows the two to bypass the middlemen. I had some reservations about one of the aspects of the goCatch design as described by James, i.e., its statefulness, which led to problems of synchronisation of state held on devices with that on the server. Perhaps a suitable set of messages based on idempotence could solve the problem. I didn't have time to discuss this offline with James.
Our fourth and final talk of the evening was by Darren Younger, CTO of IPScape, in whose offices the meeting was held. IPScape is a provider of cloud-based contact centre solutions catering to both voice and web. One of their interesting applications allows mobile device users to make phone calls not through the device's native telephony capabilities, but through the IPScape app. The server then initiates regular (teleconference-style) phone contact with both the caller and the receiver. The advantage of this is that the server can record the call. Many financial service providers are required by law to record all customer conversations, and it is easier for them to use this app rather than approach the telcos for a voice recording service. A developer API may be coming in a few months.
I also met another attendee, Nanik Tolaram, an amateur Android enthusiast with his own Android-related website.
I picked up a few useful tidbits of information over the course of the evening.
Samsung sent a couple of people to the meeting. They seem keen to understand the size and strength of the Android developer community. Samsung wants to carve out a unique niche even within the Android ecosystem and they have their own app store separate from the generic Android one.
DroidDraw is a testing tool for Android User Interfaces.
Google has a cloud-to-device API.
developer.android.com is Google's portal for Android developers.
Balsamiq is a commercial tool to mock up UIs, including mobile UIs. Pencil seems to be a good Open Source equivalent, and Lumzy is a free one.
Two of the presenters talked about their negative experiences with outsourcing. Although the outsourcing countries were as varied as Israel, India and Singapore, there seem to be some common problems caused by distance as well as the seeming cultural inability of some developers to look beyond the literal specification and to understand the higher abstraction that an application is trying to implement. Errors in the documentation of some specs led to literal implementation of those features even though they patently made no sense. Outsourcing sites like Freelancer.com seem very cost-effective, but the elapsed time to obtain a working solution negates those benefits. Examples: $200 and 4.5 months to develop an app, $800 and 9 months to develop another. The moral of the story seems to be to hire good local developers so communication problems are reduced and results are achieved quickly.
The Android Australia User Group is a good place for developers to hang out. The organisations that some of the speakers represented are looking for developers, and this may be a good way to get introduced.
