Certainly. The lack of error checking and documentation made it non-obvious why
"10" segfaulted, but this is me poking at a quick prototype from 30+ post
facto.
Looking at this incunabulum as a proof of concept, it feels like I can almost
hear whispers of the discussions that happened at Kiln Farm that weekend.
Adding error checks and the like certainly would detract from that
conversation, as you point out.
I am somewhat enamored. These lines here are beautiful:
C vt[]="+{~<#,";
A(*vd[])()={0,plus,from,find,0,rsh,cat},
(*vm[])()={0,id,size,iota,box,sha,0};
What a great way to correlate the symbols with their monadic and dyadic uses.
All in three lines! I had never thought to put function pointers in a literal
array like that.
Somehow this bridged a gap between my J-brain and C-brain. I can already feel
how my imperative code will be affected.
Raul Miller <[email protected]> wrote:
> Sadly, jsoftware.com/zippy.htm is gone, so I can't refer to it here
> (and the associated gif is transparent and looks a mess on the black
> background used by modern browsers).
>
> Anyways, here's some different perspectives on the incunabulum:
>
> (1) It's a model. Adding error checking to give error messages when
> the user used unsupported syntax would have increased its size by
> several orders of magnitude. And, this would have mostly served to
> conceal the model.
>
> (2) When working with compilers which are capable of representing
> hardware features, crashing is something you need to incorporate into
> your workflow (you need coping measures for it).
>
> (3) brevity is the soul of ___
>
> Thanks,
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
