Mark Howard, director of software engineering at AXA Rosenberg, describes why we chose Eiffel for our expert system.

I'm just reading Yhc/Erlang/Proof of concept.

Perhaps some enterprising young student would take a stab at compiling Eiffel to BEAM? Then we'd get real concurrency.

Not a trivial task.

Eiffel is based (according to Wikipedia) on a set of principles:

• Design by contract
• Command-query separation
• Uniform access principle
• Single choice principle
• Open-closed principle
• Option-Operand separation

I have just released Gestalt 1.0, an XSLT 2.0 command line interpreter written in 100% pure Eiffel, on the suspecting wide world.

The released 32-bit binaries for Linux and Windows were compiled with GEC and the Boehm garbage collector (7.0). Any offers to contribute additional binaries will be accepted.

It seems to me that STRING_GENERAL isn't worth having - just take a look at some of the implementations, and you will see they are marked for 8-bit only.

I think it would be better to take the opportunity to abandon read-write strings altogether (confining STRING_8 to "legacy") and make STRING_32 unconnected with STRING_8.

Whilst watch Eiffel Studio finalizing, I saw:

Degree -2: Constructing Polymorphic Table

Q. How many legs does a polymorphic table have?

A. ?

Anyone fancy having a go at tackling http://www.tbray.org/ongoing/When/200x/2007/09/20/Wide-Finder this in Eiffel?

Routines that simply forward the request to another object (a delegate) typically have the same signature and contract (possibly strengthened) as the delegate's routine. In which case it is slightly tedious to write out the duplicate assertions (especially if the delegating routine is using different names for the arguments), and a potential maintenance problem too.

See http://colina.demon.co.uk/?q=node/31

I used to see mention of a keyword "pure", but it didn't make it into either edition of ECMA.

I would have thought that it would be extremely useful to indicate that a function is referentially transparent. Compilers could cache the results. Automatic reasoning tools would benefit. And I would think that the (eventual) verifying compiler would need it.