Re: Q: access to subprogram

[dewar@cs.nyu.edu (Robert Dewar)]

Bob Duff said

"sure, but I think that perhaps static links are faster.  I'm quite
*sure* of two things: it doesn't make much difference, since most code
isn't very deeply nested anyway, and if there *is* a difference, it's
not a big difference."

If there is no nesting, then neither static links nor displays have
any overhead at all.

If there is nesting, then static links pay a price for calling any routine
other than an outer level library routine.

If there is nesting, then displays pay a price for calling any routine
that contains a nested routine, but there is no price for calling any
routine (at library level or nested) if it contains no nested routines.

It seems to be evident that the great majority of calls are to routines
that do NOT contain  nested routines. Consider in particular the case
of a big procedure q that contains many tested procedures q1 .. q100.

Now for static links, calls to any of q1 .. q100 incur an overhead, but
the call to q does not (assuming q is global level).

For displays, calls to any of q1 .. q100 (assuming there is no further
level of nesting) incur no overhead, only the call to q incurs an overhead.

It seems clear to me from these considerations that displays will be
faster than static links for most typical programs.