RE: accumulator generator (Java)

["Anton van Straaten" <address@hidden>]

> It isn't a test of "power".

It isn't if you don't accept the premise "succinctness = power".

> Now if you showed a circumstance (and there are many) where hundreds of
> lines of hard-to-read, hard-to-maintain Java collapse into a few lines
> in a dynamically typed programming language then even a Java programmer
> would probably admit that they've traded some expressive power for their
> safety.

The question is whether they really have to trade in that expressive power,
or if the restriction arises from limitations in the language that other
languages might handle in a more expressive way, while still being typesafe.
An example might be a type inferencing language - see the O'Caml example
just posted.  So once again, I see value in this microbenchmark.

In Java's case, a trumpcard that can be used to justify many of its
restrictions is that of security, including dealing with untrusted code and
so forth.  That's a tough one - talk about a "crosscutting concern".  That
concern may indeed justify a great loss of expressive power.  But I don't
think type safety alone, or most other "ordinary" language features, really
has to result in loss of expressivity.

Anton