guile-commits
[Top][All Lists]
Advanced
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Guile-commits] GNU Guile branch, master, updated. release_1-9-0-20-gb67

From: Andy Wingo
Subject: [Guile-commits] GNU Guile branch, master, updated. release_1-9-0-20-gb674d47
Date: Mon, 22 Jun 2009 20:44:50 +0000 
This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "GNU Guile".

http://git.savannah.gnu.org/cgit/guile.git/commit/?id=b674d4716abe775e648445795f02bece8a3396e0

The branch, master has been updated
       via  b674d4716abe775e648445795f02bece8a3396e0 (commit)
      from  eb72179985966493f452c7be8b9b048341d2f9c5 (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
commit b674d4716abe775e648445795f02bece8a3396e0
Author: Andy Wingo <address@hidden>
Date:   Mon Jun 22 22:44:34 2009 +0200

    more docs to brainfuck->tree-il compiler
    
    * module/language/brainfuck/compile-tree-il.scm (compile-tree-il): Wrap
      the result in a ((lambda () ...)), so we can use toplevel-ref. Add
      lots more comments.

-----------------------------------------------------------------------

Summary of changes:
 module/language/brainfuck/compile-tree-il.scm |   63 ++++++++++++++++++++----
 1 files changed, 52 insertions(+), 11 deletions(-)

diff --git a/module/language/brainfuck/compile-tree-il.scm 
b/module/language/brainfuck/compile-tree-il.scm
index c991631..62987fc 100644
--- a/module/language/brainfuck/compile-tree-il.scm
+++ b/module/language/brainfuck/compile-tree-il.scm
@@ -49,32 +49,66 @@
 ;; This compiles a whole brainfuck program. This constructs a Tree-IL
 ;; code equivalent to Scheme code like this:
 ;;
-;; (let ((pointer 0)
-;;       (tape (make-vector tape-size 0)))
-;;   (begin
-;;     <body>
-;;     (write-char #\newline)))
+;; ((lambda ()
+;;    (let ((pointer 0)
+;;          (tape (make-vector tape-size 0)))
+;;      (begin
+;;       <body>
+;;       (write-char #\newline)))))
 ;;
 ;; So first the pointer and tape variables are set up correctly, then the
 ;; program's body is executed in this context, and finally we output an
 ;; additional newline character in case the program does not output one.
 ;;
+;; The fact that we are compiling to Guile primitives gives this
+;; implementation a number of interesting characteristics. First, the
+;; values of the tape cells do not underflow or overflow. We could make
+;; them do otherwise via compiling calls to "modulo" at certain points.
+;;
+;; In addition, tape overruns or underruns will be detected, and will
+;; throw an error, whereas a number of Brainfuck compilers do not detect
+;; this.
+;;
+;; We wrap the code in a lambda so that the body has a place to cache
+;; the looked-up locations of the primitive functions: vector-ref et al.
+;; This way we can use toplevel-ref instead of link-now + variable-ref.
+;; See the VM documentation for more info on those instructions.
+;;
+;; Normally when compiling you don't have to think about this at all,
+;; because the usual pattern is a bunch of definitions, then you call
+;; those definitions -- so the real work is in the functions anyway,
+;; which can use toplevel-ref. Here we just force that pattern into
+;; effect.
+;;
 ;; Note that we're generating the S-expression representation of
 ;; Tree-IL, then using parse-tree-il to turn it into the actual Tree-IL
 ;; data structures. This makes the compiler more pleasant to look at,
 ;; but we do lose is the ability to propagate source information. Since
 ;; Brainfuck is so obtuse anyway, this shouldn't matter ;-)
 ;;
-;; TODO: Find out and explain the details about env, the three return values 
and
-;; how to use the options.  Implement options to set the tape-size, maybe.
+;; `compile-tree-il' takes as its input the read expression, the
+;; environment, and some compile options. It returns the compiled
+;; expression, the environment appropriate for the next pass of the
+;; compiler -- in our case, just the environment unchanged -- and the
+;; continuation environment.
+;;
+;; The normal use of a continuation environment is if compiling one
+;; expression changes the environment, and that changed environment
+;; should be passed to the next compiled expression -- for example,
+;; changing the current module. But Brainfuck is incapable of that, so
+;; for us, the continuation environment is just the same environment we
+;; got in.
+;;
+;; FIXME: perhaps use options or the env to set the tape-size?
 
 (define (compile-tree-il exp env opts)
   (values
    (parse-tree-il
-    `(let (pointer tape) (pointer tape)
-          ((const 0)
-           (apply (primitive make-vector) (const ,tape-size) (const 0)))
-       ,(compile-body exp)))
+    `(apply (lambda () ()
+              (let (pointer tape) (pointer tape)
+                   ((const 0)
+                    (apply (primitive make-vector) (const ,tape-size) (const 
0)))
+                   ,(compile-body exp)))))
    env
    env))
 
@@ -137,6 +171,13 @@
         ;;         (begin
         ;;          <body>
         ;;          (iterate))))
+        ;;
+        ;; Indeed, letrec is the only way we have to loop in Tree-IL.
+        ;; Note that this does not mean that the closure must actually
+        ;; be created; later passes can compile tail-recursive letrec
+        ;; calls into inline code with gotos. Admittedly, that part of
+        ;; the compiler is not yet in place, but it will be, and in the
+        ;; meantime the code is still reasonably efficient.
         ((<bf-loop> . ,body)
          (let ((iterate (gensym)))
            (emit `(letrec (iterate) (,iterate)


hooks/post-receive
-- 
GNU Guile
reply via email to
[Prev in Thread] Current Thread [Next in Thread]