[racket] eginner's question on elementary textual replacement...

From: Matthew Flatt (mflatt at cs.utah.edu)
Date: Mon Mar 12 11:00:53 EDT 2012

If you're interested in the compilation that happens after macro
expansion, including constant-propagation and inlining, then try `raco
make' and `raco decompile'.

For example, if "alias.rkt" has

 (module alias racket
   (provide plusone)
   (define ONE 1)
   (define (plusone x) (+ x ONE))
   (plusone 2))

then `raco make alias.rkt' followed by `raco decompile alias.rkt' shows

(begin
  (module alias ....
    (define-values (_ONE) '1)
    (define-values
     (_plusone)
     (begin
       '%%inline-variant%%
       (#%closed
        plusone9
        (lambda (arg0-10)
          '#(plusone #<path:/private/tmp/alias.rkt> 4 3 65 30 #f)
          '(flags: preserves-marks single-result)
          (#%in + arg0-10 '1)))
       (#%closed
        plusone8
        (lambda (arg0-13)
          '#(plusone #<path:/private/tmp/alias.rkt> 4 3 65 30 #f)
          '(flags: preserves-marks single-result)
          (#%in + arg0-13 '1)))))
    (#%apply-values |[email protected](lib "racket/private/modbeg.rkt")| '3)))

There's a lot of noise in the output, but the main result is the `3' in
the last line that is printed as the result value --- which shows that
`(plusone 2)' was reduced to `3' at compile time.

At Sun, 11 Mar 2012 16:17:43 -0500, Robby Findler wrote:
> Yes, that is correct. Macro expansion happens before any of the
> optimization steps performed by the compiler.
> 
> Robby
> 
> On Sun, Mar 11, 2012 at 4:06 PM, Thomas Chust <chust at web.de> wrote:
> > On Sun, 2012-03-11 at 22:00 +0100, Rüdiger Asche wrote:
> >> [...]
> >> So what made you think that defines within modules are inlined? Is it a doc
> >> bug, or were you looking at something else than liberal expansion that needs
> >> additional work? What does it take for define to translate into
> >> define-syntax within a liberal expansion context?
> >> [...]
> >
> > Hello,
> >
> > section 18 of the Racket guide suggests to me that module local bindings
> > which are not exported, not mutated and have sufficiently simple values
> > will be inlined. This is something that will happen in the bytecode
> > compilation or JIT compilation stages, after macro expansion has already
> > taken place, so I don't think you will see it in the macro stepper.
> >
> > However, I'm no expert concerning the internals of Racket's compiler and
> > I may be mistaken here.
> >
> > Ciao,
> > Thomas
> >
> >
> > --
> > When C++ is your hammer, every problem looks like your thumb.
> >
> >
> > ____________________
> >  Racket Users list:
> >  http://lists.racket-lang.org/users
> 
> ____________________
>   Racket Users list:
>   http://lists.racket-lang.org/users


Posted on the users mailing list.