It has happened at least once. The company went out of business in the
90th. The company was Lucid (yes, of the EMACS schism fame), the product
was Energize.

Some say they actually made money with Energize. I don't know. I touched
Energize once and couldn't stand it.

/Thomas

I dont think it's meant in a way that the IDE replaces the language. I think
what is suggested is due to the fact that the logic (code) is stored in a
database in a certain format the IDE can provide more extensive
visualisation and input options, specifically more context sensitivity. But
perhaps i missed the point of the project to begin with, a spreadsheet based
approach appeals to me more than a database based approach in any case.

Either way, i dont really see this taking off since i fail to see any
significant wins when it comes to software development. Implementing logic
(writing code) is currently only a small part of the design and
implementation of a software project and any wins there in terms of
efficiency and ease of use wont really change the world of computer
programming. I see more future in moving to a language of which programs can
be "proven" to follow a pre-defined specification of it's functionality
(instead of beating the life out of a program by unit testing it ;), as well
as standardized visualisation solutions for such programs. Consider formal
specification languages such as Z. In the case of Z a mathematical notation
is used to formulate such specifications, but i can see that moving a syntax
that's more in tune with what the average developer is used to although i do
believe at some point we'll have to move away from text-based storage for
our "code".

On a shorter term it might be possible to develop a language that's more
strictly defined than Java. I dont remember the name but i once read about a
language for which the need of threads and synchronization was removed
because the language was defined in such a way that the run-time environment
could safely decide how to have the program run on seperate threads etc. It
maybe fun and worthwhile to have a few volunteers to come up with such a
specification that's as closely related to the Java language as possible,
for the hobbyists amongst us ;)

Moral of the story is, my vote would go to a future where programming
becomes more about defining intended behaviour rather than define logic and
work towards intended behaviour. If that makes sense (sorry english isnt my
native language).

My $0.02

On Tue, 04 Apr 2006 16:58:54 GMT, Jeffrey Schwab
<jeff@schwabcenter.com> wrote, quoted or indirectly quoted someone who
said :

several reasons:

1. it requires a new way of thinking about code. People resist change.
They want to continue doing everything exactly as before. I think IDEs
are gradually softening people up for the transition.  Instead of
composing a free form text file, you are building a heavily
cross-checked data structure.

2. It needs to be introduced simultaneously with dynamic version
control.  Version control is so malodourous at its current stage of
evolution, that it will be the carrot to get people to switch.

3. egotistical programming.  People who write code think of it  as
belonging to them.  They want to pee on it -- format it in ways that
make it clear to others this is their territory.  They don't like the
idea of their masterwork being massaged in reused in a thousand ways
that remove their scent from it.  Even something as simple as whether
concatenation operator + shows up in a different colour, the original
author wants to control. If he does not like the idea, he does not
want anyone else using it.  It is HIS masterwork after all.

4. Programmers are NOT interested in productivity, unless they happen
to also own companies. They happily tell other professions how to be
more productive, but they strongly resist applying the same techniques
to themselves.  We are  much like furniture craftsmen of old decrying
the factories mass producing furniture.  Though there will always be a
market for hand-crafted Stradivarius quality software, the mass market
is much better served by something more easily cranked out to fit the
budget.  I am not calling for the equivalent of McDonald McCrap meals,
but the uniformity and much more formal quality control that new tools
could bring.  

5. My basic idea is to explain the facts of life to a computer only
once, do it properly, and ever after use that rather than creating a
million variants.  Sun is beginning to understand this concept with
the Calendar and Collection classes. They still have no business
classes that automatically validate phone numbers, addresses, zip
codes, email addresses, URLs etc, automatically refreshing the rules
from a central source.

The longer we resist improving productivity, the folk in India will
outprogram us, even if they use old fashioned techniques. They simply
work for less money, same as they outcompete us at making handmade
baskets. To compete, we must work smarter so that software creation is
not so labour intensive and is more about knowing your customer and
anticipating or fathoming his needs

Exactly, (all?) variables would require constraints (min/max bounds or more
specific value ranges, not null), also, more specific types would be
required (in other words, have type "index" for an index integer type that
always has min/max bounds..so not abuse "int" for everything). Non-nullable
variables can also help to indicate problems at compile time. There also
other small changes "int i = Integer.MAX_VALUE + 10;" should not compile,
and even in our imaginary language (i didnt really put much thought in
defining bounds) :

int a<10..20> = 15;
int b<0..40> = 20;
int c<0..30> = a + b; // <- shouldnt compile obviously

or

MyObject a = someObject;
MyObject<notnull> b = a; // Shouldnt compile, a isnt nonnull defined and may
be null