The paradigm of containers and self-referenced names

[Mark Wieder]

Mikey-

Friday, March 25, 2005, 7:12:26 PM, you wrote:

M> Implicit, implied typing is a tremendous leap in programming
M> languages.  I know what I mean, and the machine is my (uh, five letter

...and I'm in agreement there. However...

there are occasions when I really need strong variable typing. Like,
for instance, if I'm trying to pass two floating point numbers, a
string, and a date to an external function. Do I really have to go
through the C pain of converting and casting the strings I get from
the runrev engine? And for each argument? That's the main thing that
stands in the way of a generic shim to external function libraries.

I don't want variable typing most of the time. But when I want to turn
it on, I want it to be there for me.

I've gone the other way, too. Forth is about as close to the metal as
you can get. The RCA 1802 processor (this dates me, I know) didn't
have subroutine opcodes, so you had to take an 8-bit value to use as
an address, push it on top of a stack, then pop it off into the
program counter. Made it a perfect forth machine, but it couldn't be
done if there were any such thing as variable typing. Could get in a
lot of trouble that way, too. But I digress...

M> are typed (i.e. non-locals are typed), and Type All Variables.  Mark
M> might like 4D better than RR.

No... I ran away from 4D some time ago. About the time I figured out
that there really *wasn't* any documentation. And ranted about this to
the 4D folks at MacWorld and was told that that's the way it is -
you're supposed to take our courses to find out how to program it. How
long do you think RunRev would last with an attitude like that?
Frankly, I'm surprised 4D is still around. I don't mind programming in
Pascal, but at least let me know what functions are available and what
arguments they might take...

-- 
-Mark Wieder
 mwieder at ahsoftware.net