[erlang-questions] "Erlang as a First Language" -- crazy? or just stupid?
Tue Dec 29 17:36:48 CET 2009
It seems to me that the system being proposed should strive to hide as much
of the imperative code as possible, regardless of the language used. If
you can build a framework that would allow the end user to write essentially
"everything" as declarations, then the value of an IDE or source level
debugger are reduced.
The frameworks that I have worked on are in the business/financial domain,
and I find that constructing a framework that allows the business rules to
be written as predicates or patterns that can simply be "plugged-in" to the
framework is the best approach.
I would start by determining whether this approach is sufficient for your
needs. If it is, I think the IDE/debugger debate become moot, and it will
drastically reduce the learning curve for your "non-programmer" users. And
if this declarative approach does work for you, you'd be better selecting a
language that is more declarative (such as a functional language). This can
be done in a an OO/imperative language such as Java, but this would
typically mean creating a class for every rule that implements some common
interface, which could mean a lot of boiler-plate.
If you are disposed to provide more deails about the projects requirements
and goals, I'd be happy to elaborate.
More information about the erlang-questions