🚀 go-pugleaf

In article <ROCKWELL.92Apr10233251@socrates.umd.edu>, rockwell@socrates.umd.edu (Raul Deluth Miller-Rockwell) writes:
|> [...]
|> And, I don't know about you, but to me it seems kind of silly to
|> introduce a "safety mechanism" that entirely ignores the big picture.

With respect to large scientific/numerical programs, this safety
mechanism _IS_ the big picture - or a large part of it anyway. 
It seems silly to me to introduce automatic generality to arbitrary
rank arrays or arbitrary data types when such generality ignores 
the big picture - which is that few algorithms are stable (or even
possible) in such generality.  An algorithm which is fast and
numerically stable on 2-d meshes may not even generate the right 
_sign_ bit on 3-d meshes - or may be _very_ slow to converge.
In such a regime, J ignores the important points and provides
irrelevant features.

J. Giles