Quick, what does the i stand for in "for (i=0; i<n; i++) {"?

It was a quick prototype written by one APLer, for another APLer, using mutually understood conventions. For example, its dyadic verbs ("V2" functions) use the variable names 'a' and 'w'. For a while, I wondered why (maybe they stood for "Arthur" and "Whitney"?), but after reading Iverson's _A Programming Language_, I realized APL usually uses lowercase alpha (α) and omega (ω), which look like 'a' and 'w' in ASCII. Details like that would need no explanation.

Likewise, in

    typedef struct a{I t,r,d[3],p[2];}*A;

t is type (is it boxed? 0 or 1), r is rank (number of dimensions), d is the dimensions (hardcoded to a max of 3; a 2x3x4 matrix has d={2,3,4}), and p is a variable-length array sized by malloc-ing extra space; that's where the values go, either longs or void pointers). That'll save you some time if you ever try to pick the rest apart. :)

APL code is often terse to the point of seeming cryptic, just math papers are. You get used to it, and it's that way for a reason.