Lambda the Ultimate

The moment we start believing that there is only one right language, operating system, modeling language, or computational metaphor we are in serious trouble.

It is indeed unfortunate that important ‘languages’ such as Java, UML and XML, have, through their commercial success, accidentally become ”weed languages” that choke out other interesting language varieties that are perceived as less industrially relevant or worse, without utility. Instead of only focusing on the use of the latest “right” technology, a complete education in computer science should include some basic principles which will allow the student to contribute in a future work environment.

Beginning in the early 1990s, much of the diversity in programming languages, programming methodologies, different computing paradigms, and so forth, disappeared. Why? Probably because object-oriented languages, methodologies, designs, and architectures became so commercially successful...

And now, since we've been doing objects for about ten years with pretty good success, we don't look at functional languages, or logic-based languages, or declarative languages -- we don't look at lots of different things that we used to.

The upshot is that currently popular languages (such as Java) are abysmally bad at doing concurrent programming. Non-mainstream languages such as Erlang and Oz are much better at it. (By the way, there is recent work on an extension of Java called Flow Java that adds dataflow concurrency, Oz style, to Java in a very nice way.)

Concurrency is just one example among many. The point I want to make is that there is a lot of room for improvement with respect to currently popular languages. If you think that current languages are the end of history, then think again! In 1980, a Byte magazine article stated: "There are a million CP/M users. Therefore CP/M will never die!". Who is still using it today?

Heck, they are not even the beginning of the end of history...

From the outside it's difficult to assess how much todays languages dominate CS Education. There are comments which suggest monoculture and there are comments which suggest diversity.

Of course, what I really care about is the opportunity to work with more powerful tools.

where?! i know of the csp java stuff from peter welch et al, but i presume this is different - do you have any references? i did try googling, but failed miserably (there seems to be a class called Flow.java in sun's code base...!)

Flow Java: Declarative Concurrency for Java

(thanks for the reference - i've emailed schulte)

however, i presume i can explain the main point, which would have been covered by his talk today. the work extends java (compiler and vm, as far as i can tell) by adding single assignment variables (an additional qualifier to variable declarations - "single"). once you do that, everything carries across from oz, more or less (i presume).

so i'm kind-of sad - i suspect single assignment has a snowball in hell's chance of being accepted into the main language if it implies changes in the vm.