🚀 go-pugleaf

In article <1993Jan22.082118.21781@cc.umontreal.ca> kardank@ERE.UMontreal.CA (Kardan Kaveh) writes:

|
|   From: kardank@ERE.UMontreal.CA (Kardan Kaveh)
|   Newsgroups: comp.lang.scheme.c
|   Date: 22 Jan 93 08:21:18 GMT
|
|   What's wrong with the following?  MIT Scheme 7.2 complains f1 is unbound.
|
|   (define (f1 x) (* x x))
|
|   (define-macro (m1 y) `(+ 2 ,(f1 y)))
|
|   (m1 3)
|
|   ;Unbound variable: f1
|

The environment where the macro expander procedure 
 (i.e. (lambda (y) `(+ ... )) ) is closed is _not_ the top level
environment where F1 is bound.  It is a separate environment.

This may seem like a strange decision, but imagine that F1, M1, and a
use of M1 are in some file that is to be compiled.  At the time of
compilation, M1 needs to be expanded, but the expansion requires the
binding for F1 that will not exist until run time.  Since clearly,
when compiling, the macro expander cannot be closed in the runtime
environment, we decided to make a special environment for macro
expanders to be closed, and this is done consistently when compiling
and simply interacting.

The environment where macro expanders is closed is the value of the
global variable SYNTAXER/DEFAULT-ENVIRONMENT.
In your init file, you can include

    (set! syntaxer/default-environment user-initial-environment)

or some such, and then it will have the behavior that you want when
interacting.

|
|   Also, I can't seem to find any documentation on macros in the info files.
|

Correct.  There is a very short document giving an overview (and
pointing out some gotchas) but it is not integrated into the info
structure.

You can get it from altdorf.ai.mit.edu:archive/scheme-7.2/macros.txt
by anonymous ftp.