"The early versions of Eclipse weren't that good," he recalled, "and we originally saw it as a competing product. But after a couple of years of conversation inside Borland - and with the release of Eclipse 3, with its clean plug-in architecture - we had that 'ah-ha' moment. Here was this set of general purpose, open-source frameworks that actually freed us to focus our efforts on value built on top."
JBuilder 2007 was the first version of the IDE built on the Eclipse framework, in 2006.
While Eclipse was creating a forced paradigm shift among software development toolmakers, Intersimone was considering a shift in the development paradigm itself.
"In the early days, programming was just about the languages, high-speed compilers, and debuggers," he said. "Then it was about the frameworks - the ObjectWindows and MFC [Microsoft Foundation Class] libraries, user-interface frameworks, database frameworks. Then it was components. Now it's application modules and higher levels of abstractions, not just in fundamental data types and operating system components, but applications as well."
And that leads to what Intersimone calls "application-driven development." This model, he explained, pulls the structure, evolution, and logic behind developing an application into the application itself. The result is reusable software modules that capture the intent of the coder in annotated templates or "cheat sheets" that stay with the architecture of the applications.
Intersimone calls these code-plus-metadata modules "application factories."
Application factories debuted in JBuilder 2008, released in April.
"Say you need to build a shopping cart, or an ecommerce system, or an ERP system," Intersimone said. "All of those things have been built before, so if the tooling can help make that code into a reusable module - into an app factory - then you don't have to think about the infrastructure. All you have to think about is the kind of application you need to build. People do this all the time in consulting gigs, or in the large system integrators. If they build a financial system for one bank, they build it for a hundred banks."
"We think of that as tooling, not for the IDE, but for the application," Intersimone said. "In application factories, the programmers can do tooling for their applications, so they can provide simpler interfaces for making changes at the application level and to the application code."
The metadata captured in application factories would also help with the maintenance of existing applications, Intersimone said. "The everyday things that can be captured by the person who is producing these modules can free other developers from wasting time finding all the places where they have to make changes, or figuring out what the person did the last time they touched an application."
Unfortunately for Intersimone's optimism and enthusiasm, CodeGear's lack of resources showed and JBuilder 2008 was rough around the edges.
Once again, with feeling
Intersimone appears to believe things will be different under the new owner, Embarcadero. His blog makes repeated references to the company's combined staff of 500 and revenues of $100m, which could help to build and deliver more innovation and more polish than the unwanted CodeGear child of Borland was able to achieve.
And, as with the old days of Borland when Microsoft was the main competitor on Windows, the ethos remains as ever: platform independence. This time, it's independence at the database layer. Also, there's the continued belief in vendors adding innovation on top of open source frameworks, rather than leaving it up to the community to come up with something wonderful on its own.
"Some people would tell you that tools don't matter. That open source or the platform and stack vendors will solve every problem and satisfy every pain point," Intersimone blogged after we spoke to him.
"Today, software is no more reliable, no more predictable than it was before. We need to build better applications, faster, and with higher reliability. Our economy needs every designer, architect, developer, and administrator to be performing at the top of their game".®
Additional reporting by Gavin Clarke
Goes to credibility... again....
"a reinvented version of its venerable Turbo line called Developer Studio"
Wrong. It was in fact the other way around. The Turbo product line was dead. What CodeGear did was re-invent their Developer Studio product as a suite of personalities then released cut down single personality editions as an all new "Turbo".
That was then. Now however, those Turbo's are pretty much Dodo's too, being based on Dev Studio 2006, with 2007 out in the wild for over a year and 2008 just around the corner.
If the rest of the article is more accurate, it doesn't bode well for Delphi's future under Embarcadero.
Software development is the same as it ever was. Sure the buzz words are different but the problems ARE all still the same ones we've always had and always will. The promises from the "high end" tools vendors are also the same as they ever were.
Application Driven Developement, capturing intent, modelling, lifecycle.... didn't BORLAND try to go this route?
CodeGear (and it's new owners) need to wake up FAST to the fact that the development community is not comprised solely of gullible newbies who will believe that "Magic Happens(tm)" will cut their development and maintenance costs and improve software quality and without requiring any intelligence, technical nouse or experience on the part of the business analyst driving the tools.
Borland made their name by cutting through the BS and coming up with tools that got straight to the getting to the cruxof the problem - cutting code.
CodeGear need to do the same.
microsoft played dirty against netscape?
so creating a browser that was faster, used less memory, crashed less and didnt suffer things like all the colours vanishing because you resized the window(*) is playing dirty?
A lot has changed
I think DavidI and his band of merry men are on the wrong track. Many Delphi customers have been complaining about the lack of a Win64 version for several years now. CodeGear Delphi now only covers a small part of the Windows platform surface. They need to focus on the things that only they -- the tool vendor -- can do. Nobody else can change the compiler. Nobody else can change the core component library to fit in better with new Windows versions. And where is the unicode support they should've put in ten years ago?
Too much stuff exists only on a dusty roadmap and in competitor's products, rather than installed on my hard drive.
Not everyone is interested in UML and application factories. Some of us still hand code shrinkwrapped Windows software and we need support for the OS. .net support seems like a blind alley, because most of us will simply use Anders' quite adequate C# language for that. OK, some enjoy eco.net, but I suspect those are in a minority (eco.net deserves more attention, I am sure, but big companies with big apps love to waste money on big development teams, and they rarely bother with saving time nor cost).
Free Pascal seems to be heading in the right direction though, so there is hope yet. For an object oriented language that is. Hopefully CodeGear will pick up the slack soon.