Re: Add nested-function support in a language the based on a stack-machine

On 2018-03-01, Kaz Kylheku <217-679-0842@kylheku.com> wrote:
> John:
> [A display is an array of pointers to the frames of the enclosing
> scopes. You can access any variable in an enclosing frame by
> indirecting through the frame's entry in the current display and then
> indexing. This is for languges like Algol and Pascal, not Lisp. I


This is absolutely relevant to Lisp dialects with lexical scoping,
like Common Lisp and Scheme (or anything worth its salt, really).
There are no first-class functions without lexical scoping.


Structure and Interpretation of Computer Programs (SICP) discusses
something similar to displays in "5.5.6 Lexical Addressing":


https://mitpress.mit.edu/sicp/full-text/book/book-Z-H-35.html#%_sec_5.5.6


"We can exploit this fact by inventing a new kind of variable-lookup
operation, lexical-address-lookup, that takes as arguments an
environment and a lexical address that consists of two numbers: a frame
number, which specifies how many frames to pass over, and a displacement
number, which specifies how many variables to pass over in that frame.
Lexical-address-lookup will produce the value of the variable stored at
that lexical address relative to the current environment."


The focus of that section is eliminating the run-time search which
actually finds a name by symbol matching, with numeric indexing.
However, if we can "pass over" the frames by indexing through a vector
of them, then we have a display. We can also directly index into the
frame by also representing it as a vector, so we don't have to
"pass over" n-1 variables to get to the n-th one.


[I was thinking of closures when I said not Lisp. I suppose you could
use displays and garbage collect them but I haven't really thought
about the implications. -John]

Return to the comp.compilers page.
Search the comp.compilers archives again.