I have been going crazy thinking about a solution for our new intranet which I am in charge of. I am definitely the biggest proponent of AJAX/DHTML front end, PHP/Apache/Mysql backend for most web design but with the intranet we are going to need a lot of mashup like functionality and have many different programmers working on modules that may often reside on the same page. At the same time we are being told we are never going to move past IE 6, at least not any time in the near future. But I can't base the entire intranet on IE6's capability (its so 2001) and I know eventually it will go away. Hopefully, before the intranet requires redesign.
I have no issues with the backend part. In fact now that all our new servers arrived and have been set up, I cannot be happier with that. But with IE6 being the main output mechanism for the data, I am definitely unhappy with the front end.
Not only that but it al works slightly slightly different in each browser and that I have to use a rather large javascript library to deal with those inconsistencies.
We also have the need for some desktop applications that should be build cross platform and tie in with the intranet data. Those could be developed in java but it would be ideal to have a solution that is develop once deploy in multiple formats and apple and java are doing so well together right now.
Flex 3
So this weekend I am evaluating Adobe Flex (web/air app) front end with PHP backend and it seems like a possible solution. As Zend calls it "Party in the front, business in the back."
I am not suggesting the entire intranet reside inside flash, only that modules/mini apps which are embedded in it are. The two major drawbacks that usually are with flash are slow download times and lack of the latest flash player - both not an issue at all in the context of our intranet at work.
The Flex IDE is based on eclipse which has tons of free mods for it. So is the new Zend studio neon, so is Aptana - so it will be nicce to have one development environment for evertything.
With Flex/Flash I can ouput to swf for in browser embedding or I can output to native desktop apps. Even the native desktop apps exported to adobe air have the ability to use HTML/javascript/actionscript/flash. The inbuilt browser is based on webkit, the parent of safari which means no matter what platform I targeted the code would be the same.
Here is a skeleton of what it will look like. On the right are three tabs that will load in different flash/flex based modules based on who you are. Right now I have repeating flash based vector clocks loading just to test if they were memory leaking at all and how fast it loads them. It all seems pretty good right now. Mind ou the deign will be much more refined this is just layout the areas for content. The rest of the page is DHTML/ajax

I will keep testing it this weekend but here are my feelings now:
Pros
0. I do not have to worry about the other programmer's javascript colliding in any way when multiple modules are on the page at the same time. This is huge - in the other approach with only DHTML/ajax we would have to test every single modules with every single other module of which there could easily be hundreds. I mean we would use namespacing etc, but al lot of the people developing this stuff are new to javascript and the potential for messing up is huge.
1. Flex has a 60 day trial for the pro version so we can test it out extensively with no overhead, although I am sure I want it.
2. Flex builder is amazing. It is like visual studio which would be a familiar interface for the other teams who would be makng modules for this but is based on eclipse and is multiplatform.
3. Because it is based on eclipse, skills used for zend studio neon could transfer. It is definitely made for programmers.
4. We would be losing nothing in terms of backend development as flex/flash is totally front end solution independent of a backend, meaning we could use php. This article

calls it "Flex and PHP - Party in the Front, Business in the Back"
5. Web modules work the same in every freakin browser. Who cares if they upgrade to IE6 or not then. This is a huge bonus.
6. Applications can exported as web swf, mobile apps, or as native system apps using adobe air. They work as native apps the same on every platform. Beats visuals studios windows only output.
7. It can be visually flashy without trashing the CPU the way javascript does.
8. It uses vectors.
9. The front end is scripted in actionsript which is essentially the same language as javascript with more robust options. The major advantage is there is no need for different scripts in different browsers.
11. It can even work with .NET as a backend. Is this a bonus, lol?
12. Linux Apahce is the best for doing this and that is what we are going with so it fits perfect with our current architecture.
13. If we want to do video/audio manipulation or capture, we already own flash media server!
14. With air you can use html/javascript or actionscript/mxml in it.
15. Adobe's MXML which makes widgeting a breeze. It has data grids, accordians, etc built in meaning we don't have to redevelop them in javascript.
16. Easy socket communication for real time data push (COMET) instead of pull (AJAX)
17. If the modules need to interact with the main app they can through ExternalInterface calls, the way surebert does.
Cons
0. Flex builder is $241 for gov't pricing - not very much compared to the retail price of like $700.
1. I am master of cross browser DHTML/AJAX and know the ins and out of every browsers. I even developed my own extensive library for dealing with this

- not so masterful of Flex as I am new to it. I doubt, however, that I will have much trouble.
2. Flash modules wouldn't work on the iphone. Boo, fucking boo hoo. I think that Adobe is going to address this at some point. If not I really don't care as we are using Windows Mobile at work and there is a flash 7 client already and a flash lite 3 client for that coming out. Because the backend is agnostic of the fron end, we could always make special apps for the iphone.
IMHO, Symbian's not a very good API, and neither is Gtk which, from my limited understanding of Maemo, Maemo is based on. The only surprising thing about all the other non-Qt departments getting phased out at Nokia was that it has taken as long as it has for this to happen.
Hopefully they'll start making some Qt-based tablets soon.
When I first read the subject line, I thought to myself "no oh, something happened to (e:paul) 's not phone."
I would suspect that this has something to do with the Trolltech folks. Over the years, I've heard that they don't play well with others.