Look what Phil is up to... he's got himself involved with a dynamic-oriented company unafraid to think out of the box.
Update: Related or coincidence?
Edsger W. Dijkstra with his typical insight and phrasing (he is missed)...
For the sake of the stability of the enterprise, the ideal of its manager is an organization that is as independent as possible of specific abilities of individual employees. The predominance of this ideal is a well-documented, international phenomenon; the ideal itself predates the high-technology industry, in which it could very well be inappropriate, and has discouraged the industrial eployment of scientists, in particular of the brilliant and original ones.
The American situation is aggrevated by a total lack of faith in its educational system and a deep-rooted mistrust of intellectuals...
Accordingly, prevailling industrial attitude exerts a strong pressure on the University not to indulge in such hobbies as scientific education, but to confine itself to vocational training of some sort or another.
Bill documents the growing interest in simpler languages and wonders about the right label for them.
I've always liked the term "dynamic programming language".
Mark and others have more thoughts on which analogous architecture can claim victory by correspondence with the web. I could not assign victory to Linda derived primarily from the use of
rd. This is the least unique aspect of Linda, i.e. some form of it exists in every distributed architecture.
That's from the distributed architecture perspective. Now consider the prominent developer perspective. Here the victory belongs to Tuxedo. This view is client-server with transaction management in the middle. The web server is a commodity Tuxedo. This is what Greenspun figured out, but Sun did not.
Linda can claim victory when the prominant developer perspective becomes based more on
out, and when the developers of the middle are more interested in their peers than their own rear ends.
As Jim Gray has been saying for some time now, "Memory is the new disk. Disks are the new tape." And I think enterprise applications aren't architected taking that fact into mind, and I have definite ideas how they could be.
I don't know of a naturally tree-structured language (setting aside XSLT, which isn't a general purpose language), though maybe Prolog would qualify.
Why not Lisp as a "naturally tree-structured language"?
Lists are constructed and destructured really as trees, less frequently as flat, single level lists. From list pattern matching down to the traditional list primitives (
car, cdr, caar, cadr, ..., cdddar, cddddr), Lisp and lists have been manipulating trees.
Linda - or something very Linda-like - did change the world; the World Wide Web.
This is an interesting statement from Mark, but I would not count this as a victory for Linda. Yet, anyway.
By and large, developers to date have not really used the web for automated coordination across sites.
Another thought on web continuations and congruence with other applications of continuations: Kali Scheme.
The work with Kali at NEC demonstrated how to simply implement a number of common (and uncommon) distributed programming patterns (Postscript) using serialized continuations.
The congruence is thrilling. The irony is equally (but oppositely), er, non-congruent, that this obscure concept (definition, rationale, and history) of a continuation, hanging around, awaiting its time in the spotlight.
The neat thing is these web continuations (and these) are very much like Haystack's closures/continuations and they are also like Waterken's capabilities.
These guys are really on to something. "Hijacking" a session is essentially giving someone the capability to do some work. You have the ability to pass these capabilities on to others, or not.
The ability to add "meaningful path segments" to a continuation-based URL is kind of like giving it a Pet Name.