Lambda the Ultimate

Mr. Hisao Kuroda seems to let aside an important fact - I guess intentionally. This new fashion scripting languages are all open source. Therefore he can grab the code and compile it and port it till his cadaver turns rotten.

He's also very optimistic:

ANSI Common Lisp is a dirty language. But it is standardized. In business, it's very important to be standardized. In a world of standardized language, there are many vendors and implementations, and they are compliant to the specification and interoperative.

Well, in an ideal world of standardized languages this may indeed be the case, but in the real world of standardized languages, compliance isn't a given and interoperability has to be worked into the system (even if your following the standard!). I remember back in University when students would write a program with one C++ compiler, and the professor/grader couldn't get the program to even compile with their C++ compiler. Mostly this is the result of students using non standard features, but it is possible that students are using features that are standard but haven't been implemented in the professor/grader's compiler. Now this may not be as big of a problem with some languages (Lisp or C) because their standards have been around a lot longer, but it is still an issue that one has to worry about.

You are absolutetly correct on standardization commities: standardization is as much politics (if not more) as technology.

I agree that existence of a standard does not guarantee interoperability (just remember the problems with Java memory model - well, fixed in the current version of the spec).

Also, my personal problem with specifications is that they require a lawyer to interpret them. I prefer formal semantics (which, of course, require a mathematician to interpret them, but at least semantics is much more concise). Compare a definition of pi calculus with a specification of Java^TM Business Integration. I know JBI has a much broader scope, but still - couldn't the spec include a formal semantics at least for its core functionality (exchange of messages, and conversion of them)? That would greatly improve the learning curve. Or am I reading LtU too much to believe that it is much easier to grasp a couple of pages of formal semantics as opposed to a hundred of pages of prose peppered with pieces of Java source?

"I prefer formal semantics (which, of course, require a mathematician to interpret them, but at least semantics is much more concise)."

Using formal in the CS sense: the initials of the mathematician you need to interpret them may well be PC.

I cannot possibly read language standards like you do, but I too find more comfort reading a standard. As dry as they can be, they take a lot of hard thinking to write, so they leave you with a lot fewer possibly wrong assumptions or things to chase down.

At work I've been trying to document something (non-language). The more I try to stay away from examples, the harder it gets to write and the shorter the text becomes. Examples are always a valueable addition, but if a text throws a lot of examples at me, nowadays I think the author might just be lazy.

See also Rewriting Reddit.