I'm surprised Flex doesn't have a real HTTP client built into it. You can't do REST with just GET and POST. Being dependent on the browser is pretty lame - is this driven by some sort of other requirement or some security sand box what-nots?
I guess that is what we get for hurting the web.
How are others dealing with this?
He raises some good points - concerns that I share.
We *reluctantly* looked into Flex a few months ago. We did this after trying numerous Ajax frameworks. We were not interested in a widget here or there that did some Ajax booya. We wanted to build a full blown RIA app for a very complicated system.
After suffering through severe pain in trying to get Ajax frameworks (choose any one you like) to work with Firefox, IE 5, 6, and 7 we threw our hands up.
We heard good things about Flex so we tried it.
We weighed the pros and cons of its openness (at the time it was just shared source equiv.) and decided to proceed further. Since then, they have opened up Flex - or stated the intent to under the MPL license. We were very happy to see this development.
Flash - is of course still proprietary. Mark is focused on this. He raises good points - concerns I share in a style try not to.
Adobe is showing a good direction towards opening things up further. I'm hopeful that some day Flash will be open.
In the meantime, our business has competition and we seek to beat it with the best systems we can build. We weigh the pros and cons, make our choices, and live with the consequences. Just like Google Finance and YouTube (properties owned by Mark's employer that both use Flash) do.
Here is a good video interview with Adobe product management.
The future is coming!!
It should be very interesting to see the impacts of this.
We started working with Flex a few months ago after trying every viable Ajax RIA framework. We were very reluctant to do this at first as Flex was propietary. There were just too many "pros" to using Flex over any Ajax framework, however. We estimated the cost of development and maintenance of Flex vs. Ajax to be massive. So we went with it anyway.
But now that it is OSS, dang!
Update Changed "We estimated the cost of development and maintenance of Flex & Ajax to be massive." should read: "We estimated the cost of development and maintenance of Flex vs. Ajax to be massive."
I'm with Patrick on this one.
Mark's tried and true "just leverage the web" argument doesn't seem to apply. Mark is one of my favorite bloggers and I typically agree with him.
Then again, the Adobe person's "“One word; branding”" doesn't really make any sense to me.
Flex and Apollo are compelling because they are essentially a browser with a real run-time in them. Ajax is neat and a great way to incrementally add booya to your in-production-today-built-on-Struts web app (or whatever your equiv is). But Ajax isn't a run-time and most of us don't really want to write our app N times for N different browsers/versions.
But happilly, "in the future", we have greener pastures to look forward to. We can like "do stuff" and not like "bleed to death" doing it.
And guess what? It will run on Winders, Linux, and OS X. And you will write it one time. And test it one time.
Like what Java was supposed to do but wasn't on the desktop and you only get 10 years before people say, "times up, what's next?".
We've covered a lot of ground in this phase of the project. A lot of it was fighting learning curves and getting infrastructure stood up. We spent a lot of time using Lean set based decision making on what technology we wanted to use. This was good in that we are more confident in our decision, but bad in that it cost time. The good news is we wouldn't have picked the technology we ended up with if we had used a traditional "place your bets early" approach. So it worked. Yay.
It is inevitable yet annoying that the first rev. of any piece of software has to pay the first rev. tax of standing up infrastructure, figuring out methodology, forming/storming/norming/hopefully-performing, and learning rapidly in general.
The team makes its choices (as Sarge would say) and lives with the repercussions. We made a lot of good choices luckily, but we certainly made some bad ones. The bitch of it is you never really know which choices were good and which choices were bad unless they were very right or very wrong. You are always left wondering what if we did this - what if we did that?
And so it goes - here comes the rubber match between the Space Men and the Flex Men. I am using this phrase incorrectly, but I felt like using it anyhoo, so so what!?. It really is the integration week. When we take a lot of reasonably well tested Jini/JavaSpaces code (unit, integration, and FIT) and try to integrate it with Flex code.
For better or worse, we split our small team into two - one focusing on the backend JavaSpaces side and one focusing on the Flex UI side. This was mostly due to fighting learning curves, but was also due to working styles and geography. We wanted to try a lot of things and splitting the team at the time seemed to be a way to cover the most ground with technology and methodology. We'll never know if this was a good decision. But we made our choices.
A team where everyone is skilled in all areas of the system is obviously best. But you can't just wave a wand and make this happen. You have to get there iteration by iteration choice by choice.
Wish us luck this week - the rubber match is going to hurt! :)
I spent too many hours yesterday trying to beat Cygwin and Rake into submission. All because of the horror of backslashes etc.
Cygwin mostly works, but there are always corner cases. Well Flex seems to be one of them. I am by no means a Cygwin expert - perhaps I should learn more.
Anyways, here is a tip. Don't even try to feed the Flex compiler (mxmlc) path names like /cygdrive/c/flex_2_sdk_2 or path it any config file references using this style.
You must instead do things like:
export FLEX_HOME="C:/tools/flex_sdk_2"
Anyways, no big deal of course.
This type of panic is only typically an issue when you are dealing with a new tool. This is why I love the project "Work Instructions" wiki page. This is the step by step instructions for the new developer (or you when you get a new machine) that tells you exactly what to do to get productive quick. Even better - a team starter VmWare image or some such.
As this article describes:
They [Powerbuilder programmers] could do stuff quickly, or using the modern jargon, they were agile programmers without even knowing it.
One of the smartest people I know, my co-worker Erik Onnen recently tried every Ajax framework available and tore his hair out trying to deal with browser issues (don't even think about IE 6). This is not some browser neophyte. He knows JavaScript, JSON, CSS etc. inside and out.
Every Ajax intesive site I have been on besides Gmail locks up, bombs out and as another co-worker of mind (Ed Copony) says, "Ajax apps. make your computer fan go insane".
But I want to build a RIA app! I just don't want all the panic!
4-5 years ago building Flash GUI's was the rage for a bit. I never got a chance to do it & was admittedly against it in general. But suffering through Ajax makes one take another look. It is of course called Adobe Flex today & it seems to have come a long way. Certainly seems like a *productive* way to build RIAs. Sure you have to get past a couple hang ups, but that is life in the big city.
Hmmm, what can you do with Flex + JavaSpaces + Caching ... hmmm ...
