Henry,



I offer this as tangentially responsive to your question (posted to
jprogramming), "How do you convince someone that J is really different?"



I'm an amateur J programmer and I think I know what makes J different.  I
still remember (rather hazily; see below) the moment when it hit me.  FWIW,
here is my conversion story.



One evening in a midtown Manhattan bar in 1989, over the course of nearly a
dozen pints of Guinness between us, a VP at Morgan Stanley persuaded me (a
junior Smalltalk programmer) that APL was a gift from the gods.  He told a
wonderful story about showing it to his wife (who was struggling through a
FORTRAN class): when she realized that she could just type in two lists of
numbers with a “+” between them rather than declaring, assigning and
looping, her face lit up like the very dawn.  He was bitterly convinced
that absent the unconscionable interference of Bob Frankston and Dan
Bricklin, APL—not the spreadsheet—would have conquered the world.  His name
was R0ml (yes, that is a digit) and he was without question the smartest
programmer who’s ever been willing to drink with me.



The need for a custom keyboard was enough to keep my admiration for APL
distant if sincere, but when J became available I jumped on it,
thinking,  “This
will actually be _better_ because Ken Iverson will fix everything he
doesn’t like about APL.”  I was very enthusiastic, and remained so as I
hunched over my machine, ruthlessly (if explicitly) dispatching toy
problems.  I found J (still the only array language I’ve studied) to be a
towering intellectual achievement—it was clear that far more thought had
gone into it than into any language I’d ever used.  My father (proud owner
of a threadbare “Bring Back the 7090” T-shirt) told a story about pushing
my baby carriage down the sidewalk while reading the first FORTRAN manual,
absolutely convinced that the mind of Man could do no more.  For the first
time, I knew how he’d felt.



My 30 years with J have been those of an on-again-off-again hobbyist.  Like
most "J-Adjacents," I suspect, I've tried to persuade my employer of the
moment to use it for this or that problem domain.  It always went about as
well as you might imagine.



So what did R0ml tell me that made such a huge impression?  Several
things--none surprising to this group, each a revelation to me as we drank.



   1. Most programming languages' primitives operate on scalars.  There's
   not much you can do with scalars: add them, subtract one from another, take
   a remainder, index into an array, etc.  So most programming languages
   have few primitives.



   1. You can't accomplish much with individual scalars.  You have to work
   with collections of them to get anything done.  These collections are
   often arrays.  If you find yourself looping, you're probably dealing
   with an array (if only implicitly).



   1. APL's primitives operate on arrays.  (Here he went through a few
   classic, eyebrow-raising examples.)



   1. Two implications of primitives operating on arrays: first, you
   write/test/maintain a lot less code (insert canonical example of using a
   nested loop to add two vectors).



   1. Second, because arrays are richer constructs than scalars, you can
   have a lot more primitives.  That's great news for the power of the
   language (and the length of your programs) but…



   1. …it's not the end of the story.  Learning a language means learning
   (most of) its primitives, so learning APL takes (a lot) longer than
   learning a conventional language, even before you take into account the
   geometric increase in idioms.  This higher threshold for competence
   turns out to be a significant barrier to adoption, despite APL's (array
   languages') obvious advantages once you're over the hump.



And there we called it a night (he picked up the tab).  I found the case he
made clear, compelling, horizon-shattering, persuasive
and--ultimately--somewhat discouraging.  I have yet to run across one that
I found more so.


As I said, FWIW.



Many thanks, Henry, for all of your technical- and pedagogical
contributions over the decades.



Ed