Adobe just announced that it will open source Flex under the MPL license.

I find this quite funny since when Flex was initially released you had to go through a stringent selection process before you were even granted a free developer license. At that time they judged that Bla-bla List didn't meet the requirements and my plans to create an alternative open-source version in Flex as a comparison couldn't go through.

When people have been asking me about Flex at conferences, I've been saying that to be successful it will have to be open-sourced. It seems they eventually realized that nowadays this is a requirement for any development platform. Kudos to Adobe for that.

Sadly, this will now make it even more difficult for OpenLaszlo to defend its case. I still think that it's a superior solution since the programming language is much closer to what people are used to in Javascript development. Also, the fact that OpenLaszlo now supports multiple runtime environments such as DHTML and soon JavaME, is a major bonus. However, going up against the huge Adobe marketing machine will now be any even more difficult task.

Atif Khan wrote a blog entry about his experience with RIFE. Sadly, the comment system on his blog seems to be broken, so I can't post a reply there. I'm thus posting it on my own blog.

Hi Atif,

thanks a lot for writing down your impressions, and you see that I did get to your blog entry

Here are my reactions to your 'cons':

Which template syntax did you use? If you want the designer-friendly version, you have several to choose from, like the comment based syntax and the real XML syntax. I suppose that this is a matter of being aware of them, which is not easily solved (there's a wiki page about them though).

The flowlinks and datalinks verbosity are easily solved through different manners. They're like the finest grained possibility. For a lot of people, globalvars and globalexits are totally sufficient and only require one declaration. Another good solution is just to use autolinks, if you combine these with annotations, there's very little setup to do for the logic flow and data flow to work.

Also, if you think they're not clear, this page in the users guide might make it easier to understand: http://rifers.org/wiki/display/RIFE/GuideNumberguess.

I agree that the documentation here is lacking, but annotations are a recent addition, so in time it will get better. The best thing usually is to try to combine what's on the wiki with the javadocs and the examples. For the annotations, there's some kind of overview here: http://rifers.org/wiki/display/RIFE/Best+Practices. You can also have a look at the javadocs for them.

It's a pity that you didn't see the real value of the validation framework, the MetaData classes actually generate the statements for the real validation framework underneath, which is totally customizable. Several times I've heard users say that they found RIFE's validation framework to be the most flexible and powerful one that they've come across. Information about that can be found here: http://rifers.org/wiki/display/RIFE/Validation, this also contains instructions about how to customize the error messages in many ways. The javadoc for the Validated interface might be helpful too. You can basically create any ValidationRule or ValidationError that you want, using MetaData just automates it all.

Please don't hesitate to tell me where you find holes in the documentation. I think that a lot of it is documented, it's mainly an issue of finding the right content. Shooting of a message to the mailing list will often get me or someone to point you in the right direction. There were some books in the writing, but sadly I had to cancel them due to time constraints. There might still be some coming up in half a year or so, but it really is very bound to my availability.

Again, thanks a lot for making the effort to look at RIFE, I know that it's a lot to digest.