Lambda the Ultimate

It also depends on the skills and personalities of the people involved, and on the market for people with skills in the desired language.

But it's much worse if you sometimes sell the source code for your product. Then you have to convince all your customers that Language X (not Java or C#) is so great that the hard facts that they've never heard of it, and that there are few people who know the language, are of little consequence. Sigh. Sometimes I wish for compilers that would output Java or C# that looks hand-written, so I could write in the language I want, and deliver code in the language that they want.

It's about network externalities. Finding people to work on systems, finding libraries, tools, books and so on is important. To get people to move away from that safe fold (VB, Java, C, etc...) you had better have a pretty compelling reason. They aren't just being sheep (well, all of them at least) - network effects represent real value.

No doubt these things are valuable. My point was that the value of these things may be different based on who you are and how you sell your software. If you can use a language like Nice, then you've still got access to all the network effects of Java libraries, JVM tuning & improvements, and all the Java tools that are based on parsing source code. Then there's still the question of people, books, training courses and so on, which may be irrelevant to the right team.

I'm just pointing out that if you sell your source code, you have to consider this tradeoff not just for your own team, but for all your customers as well. I might well want to use a more advanced language because I feel it could give me a competitive advantage, but not be able to because the clients just want Java source code.

So when your source code is part of your product, you're automatically deprived of the choice to use a better language. That's why I was wishing for human-style source code as a compiler target, so I could write in the language I choose, but still deliver the Java source code that the client wants.

The problem with rich web apps is ultimately their variety of interface: they become increasingly arcane, so much so that the user requires training. In contrast, navigation in a well-designed web application is apparent to a knowledgeable browser user. So in the end it's time and money spent on training and learning that will keep rich web apps out of the marketplace and that will propel the use of simple web apps.

...offers the NetSheet sample, among others.