Flex: Where are you headed?
I read an interesting post on official Flex blog few days ago. Flex: Where we are headed tries to summarise Adobe’s view on Flex framework and it’s future. Unfortunately my comment has not been approved for about two weeks now (holidays?) so I decided to post in on my blog. Hopefully some of these concerns will be addressed. Here we go:
Dear Adobe,
It’s great to read that you treat your platform seriously. You are positioning yourself as enterprise platform provider, but the reality shows you are not there yet. Several improvements to Flex and Flash need to be made for that duo to become truly useful in the enterprise world. I have worked on a number of large scale Flex projects for major financial institutions and would like to share some of my observations. I know this is an ultimate wish-list, but all these things really matter:
- Flash runtime needs to be made more efficient. Really big budgets are required to make large modular applications work smoothly. Drop some backwards compatibility baggage if you need to. Virtual machines can be really fast (JVM).
- AS3 needs to be updated (it’s 5 years old already) to have proper collections library, support for generics and decimal type. You need to make it even more expressive and powerful.
- Better reflection and code instrumentation capabilities are needed.
- Better support for modularisation and runtime dependencies is needed – make RSLs even more powerful and controllable by developers.
- General quality of code across Flex SDK needs to be improved considerably. Having so many bugs and regressions is not acceptable in mature enterprise framework.
- Performance of Flex SDK components needs to be improved. Hundreds of components and tons of data on the screen should not be a problem. That’s how enterprise applications usually look like.
- Styling engine in Flex needs to be replaced with one that is more powerful, efficient and less buggy.
- Flex compiler needs to be faster. Compile time in any larger project are so high, that may seriously hurt productivity.
- Flash Builder needs to become world class IDE – some third party IDEs are ahead (IntelliJ).
- Proper profiler is a must. Current solution is not powerful enough, doesn’t visualise results well and is way too slow.
- Better support is needed for tools that are widely used in enterprise, such as maven. Contrary to some opinions, flex-mojos is not good enough yet.
Flex developers spend too much time fighting their tools. That makes building Flex platforms rather expensive and risky investments. If this perception will get instilled in minds of managers and directors, then enterprise Flex is dead and lots of Flex developers will need to start looking for new job. Keep us informed about your plans.
Thanks.
Background thinking 
This pretty much sums up my current experience and state of mind. Been working with Flex since early beta of Flex 2 SDK and, looking at the time that has passed since then, I’m greatly disappointed in the pace of changes taking place in Flex. It’s a standing out framework, no doubt about that, but this will not last forever if Adobe won’t come back with proper solid strategy for the product.
I’m often hearing from Java developers, that they would pick up the technology if it wasn’t for anachronic and feature-poor language that AS3 currently is (lack of function overloading is often mentioned as particularly bizarre). That could be easily alleviated by putting some work into next-gen ActionScript. As for the runtime itself, facts speak against it – current javascript vm implementations in Chrome or Safari can easily beat FP. In fact, their performance improves every few months, while FP stays almost the same for couple of years now.
Personally, I would be relieved to hear Adobe dropping the mobile Flex thingy. I’m yet to hear a story from the business about successful delivery of enterprise Flex app on mobile. The idea obviously sounds great, but given the dire problems Flex already has on desktop at the moment, adding up to chaos by focusing on mobile just doesn’t look good in my book. Of course, it sounds great for marketing, but I doubt it’s something we (developers) actually *need* right now. Android world is already dominated by skilled Java folks. To transfer the skillset to just another platform, is relatively easy. iOS is hardcore, but those who made the leap are already deep into it and can work efficiently with ObjC and XCode. Doesn’t seem like the pie piece here is big enough to take even a small bite…
Maybe it has been approved since publication, but your comment can be seen on the Adobe post ;)
Yes, it showed up on the next day.