By way of followup, I coded a routine for this benchmark that uses
functions only, with the data stored in a block initialized to
[ _keys [] _vals [] ]
That block is passed as an argument to the functions that access or
modify keys and values; the functions access the two sections of the
I was interested in assessing the overhead cost of an OO version
of ADS versus a function-library version, and so wrote the
benchmarking script attached below.
It creates a block of key strings used by all of the tests (which
also all re-seed the random number generator to the same starting