gforth: gforth/prim

Diff for /gforth/prim between version 1.6 and 1.201

-version 1.6, Sat Sep 13 12:04:58 1997 UTC
+version 1.201, Mon Oct 30 15:29:48 2006 UTC
  Line 1
  Line 1
  Line 1
  \ Gforth primitives
- \ Copyright (C) 1995,1996 Free Software Foundation, Inc.
+ \ Copyright (C) 1995,1996,1997,1998,2000,2003,2004,2005 Free Software Foundation, Inc.
  \ This file is part of Gforth.
  Line 16
  Line 16
  Line 16
  \ You should have received a copy of the GNU General Public License
  \ along with this program; if not, write to the Free Software
- \ Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ \ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
  \ WARNING: This file is processed by m4. Make sure your identifiers
  Line 26
  Line 26
  Line 26
  \
  \ This file contains primitive specifications in the following format:
  \
- \ forth name    stack effect    category        [pronunciation]
+ \ forth name    ( stack effect )        category        [pronunciation]
  \ [""glossary entry""]
  \ C code
  \ [:
  \ Forth code]
  \
- \ prims2x is pedantic about tabs vs. blanks. The fields of the first
+ \ Note: Fields in brackets are optional.  Word specifications have to
- \ line of a primitive are separated by tabs, the stack items in a
+ \ be separated by at least one empty line
- \ stack effect by blanks.
  \
  \ Both pronounciation and stack items (in the stack effect) must
- \ conform to the C name syntax or the C compiler will complain.
+ \ conform to the C identifier syntax or the C compiler will complain.
- \
+ \ If you don't have a pronounciation field, the Forth name is used,
+ \ and has to conform to the C identifier syntax.
  \
  \ These specifications are automatically translated into C-code for the
  \ interpreter and into some other files. I hope that your C compiler has
  Line 54
  Line 54
  Line 54
  \ stack. Use different names on both sides of the '--', if you change a
  \ value (some stores to the stack are optimized away).
  \
+ \ For superinstructions the syntax is:
+ \
+ \ forth-name [/ c-name] = forth-name forth-name ...
  \
  \
  \ The stack variables have the following types:
- Line 72
+ Line 75
  Line 72
  Line 75
  \ df_.*         DFloat *
  \ sf_.*         SFloat *
  \ xt.*          XT
- \ wid.*         WID
  \ f83name.*     F83Name *
+ \E stack data-stack   sp Cell
+ \E stack fp-stack     fp Float
+ \E stack return-stack rp Cell
+ \E
+ \E get-current prefixes set-current
+ \E
+ \E s" Bool"             single data-stack type-prefix f
+ \E s" Char"             single data-stack type-prefix c
+ \E s" Cell"             single data-stack type-prefix n
+ \E s" Cell"             single data-stack type-prefix w
+ \E s" UCell"            single data-stack type-prefix u
+ \E s" DCell"            double data-stack type-prefix d
+ \E s" UDCell"           double data-stack type-prefix ud
+ \E s" Float"            single fp-stack   type-prefix r
+ \E s" Cell *"           single data-stack type-prefix a_
+ \E s" Char *"           single data-stack type-prefix c_
+ \E s" Float *"          single data-stack type-prefix f_
+ \E s" DFloat *"         single data-stack type-prefix df_
+ \E s" SFloat *"         single data-stack type-prefix sf_
+ \E s" Xt"               single data-stack type-prefix xt
+ \E s" struct F83Name *" single data-stack type-prefix f83name
+ \E s" struct Longname *" single data-stack type-prefix longname
+ \E
+ \E data-stack   stack-prefix S:
+ \E fp-stack     stack-prefix F:
+ \E return-stack stack-prefix R:
+ \E inst-stream  stack-prefix #
+ \E
+ \E set-current
+ \E store-optimization on
+ \E ' noop tail-nextp2 ! \ now INST_TAIL just stores, but does not jump
+ \E
+ \E include-skipped-insts on \ static superinsts include cells for components
+ \E                          \ useful for dynamic programming and
+ \E                          \ superinsts across entry points
  \
  \
  \
- Line 99
+ Line 138
  Line 99
  Line 138
  \ throw execute, cfa and NEXT1 out?
  \ macroize *ip, ip++, *ip++ (pipelining)?
+ \ Stack caching setup
+ ifdef(`STACK_CACHE_FILE', `include(STACK_CACHE_FILE)', `include(cache0.vmg)')
  \ these m4 macros would collide with identifiers
  undefine(`index')
  undefine(`shift')
+ undefine(`symbols')
- noop    --              gforth
+ \F 0 [if]
- ;
+ \ run-time routines for non-primitives.  They are defined as
+ \ primitives, because that simplifies things.
+ (docol) ( -- R:a_retaddr )      gforth-internal paren_docol
+ ""run-time routine for colon definitions""
+ #ifdef NO_IP
+ a_retaddr = next_code;
+ INST_TAIL;
+ goto **(Label *)PFA(CFA);
+ #else /* !defined(NO_IP) */
+ a_retaddr = (Cell *)IP;
+ SET_IP((Xt *)PFA(CFA));
+ #endif /* !defined(NO_IP) */
+ (docon) ( -- w )        gforth-internal paren_docon
+ ""run-time routine for constants""
+ w = *(Cell *)PFA(CFA);
+ #ifdef NO_IP
+ INST_TAIL;
+ goto *next_code;
+ #endif /* defined(NO_IP) */
+ (dovar) ( -- a_body )   gforth-internal paren_dovar
+ ""run-time routine for variables and CREATEd words""
+ a_body = PFA(CFA);
+ #ifdef NO_IP
+ INST_TAIL;
+ goto *next_code;
+ #endif /* defined(NO_IP) */
+ (douser) ( -- a_user )  gforth-internal paren_douser
+ ""run-time routine for constants""
+ a_user = (Cell *)(up+*(Cell *)PFA(CFA));
+ #ifdef NO_IP
+ INST_TAIL;
+ goto *next_code;
+ #endif /* defined(NO_IP) */
+ (dodefer) ( -- )        gforth-internal paren_dodefer
+ ""run-time routine for deferred words""
+ #ifndef NO_IP
+ ip=IP; /* undo any ip updating that may have been performed by NEXT_P0 */
+ #endif /* !defined(NO_IP) */
+ SUPER_END; /* !! probably unnecessary and may lead to measurement errors */
+ VM_JUMP(EXEC1(*(Xt *)PFA(CFA)));
+ (dofield) ( n1 -- n2 )  gforth-internal paren_field
+ ""run-time routine for fields""
+ n2 = n1 + *(Cell *)PFA(CFA);
+ #ifdef NO_IP
+ INST_TAIL;
+ goto *next_code;
+ #endif /* defined(NO_IP) */
+ (dodoes) ( -- a_body R:a_retaddr )      gforth-internal paren_dodoes
+ ""run-time routine for @code{does>}-defined words""
+ #ifdef NO_IP
+ a_retaddr = next_code;
+ a_body = PFA(CFA);
+ INST_TAIL;
+ goto **(Label *)DOES_CODE1(CFA);
+ #else /* !defined(NO_IP) */
+ a_retaddr = (Cell *)IP;
+ a_body = PFA(CFA);
+ SET_IP(DOES_CODE1(CFA));
+ #endif /* !defined(NO_IP) */
+ (does-handler) ( -- )   gforth-internal paren_does_handler
+ ""just a slot to have an encoding for the DOESJUMP,
+ which is no longer used anyway (!! eliminate this)""
+ \F [endif]
+ \g control
+ noop    ( -- )          gforth
  :
   ;
- lit     -- w            gforth
+ call    ( #a_callee -- R:a_retaddr )    new
- w = (Cell)NEXT_INST;
+ ""Call callee (a variant of docol with inline argument).""
- INC_IP(1);
+ #ifdef NO_IP
- :
+ assert(0);
-  r> dup @ swap cell+ >r ;
+ INST_TAIL;
+ JUMP(a_callee);
+ #else
+ #ifdef DEBUG
+     {
+       CFA_TO_NAME((((Cell *)a_callee)-2));
+       fprintf(stderr,"%08lx: call %08lx %.*s\n",(Cell)ip,(Cell)a_callee,
+               len,name);
+     }
+ #endif
+ a_retaddr = (Cell *)IP;
+ SET_IP((Xt *)a_callee);
+ #endif
- execute         xt --           core
+ execute ( xt -- )               core
+ ""Perform the semantics represented by the execution token, @i{xt}.""
+ #ifndef NO_IP
  ip=IP;
- IF_TOS(TOS = sp[0]);
+ #endif
- EXEC(xt);
+ SUPER_END;
+ VM_JUMP(EXEC1(xt));
- perform         a_addr --       gforth
+ perform ( a_addr -- )   gforth
- ""equivalent to @code{@ execute}""
+ ""@code{@@ execute}.""
  /* and pfe */
+ #ifndef NO_IP
  ip=IP;
- IF_TOS(TOS = sp[0]);
+ #endif
- EXEC(*(Xt *)a_addr);
+ SUPER_END;
+ VM_JUMP(EXEC1(*(Xt *)a_addr));
  :
   @ execute ;
- \+has? glocals [IF]
+ ;s      ( R:w -- )              gforth  semis
+ ""The primitive compiled by @code{EXIT}.""
+ #ifdef NO_IP
+ INST_TAIL;
+ goto *(void *)w;
+ #else
+ SET_IP((Xt *)w);
+ #endif
+ unloop  ( R:w1 R:w2 -- )        core
+ /* !! alias for 2rdrop */
+ :
+  r> rdrop rdrop >r ;
+ lit-perform     ( #a_addr -- )  new     lit_perform
+ #ifndef NO_IP
+ ip=IP;
+ #endif
+ SUPER_END;
+ VM_JUMP(EXEC1(*(Xt *)a_addr));
+ does-exec ( #a_cfa -- R:nest a_pfa )    new     does_exec
+ #ifdef NO_IP
+ /* compiled to LIT CALL by compile_prim */
+ assert(0);
+ #else
+ a_pfa = PFA(a_cfa);
+ nest = (Cell)IP;
+ #ifdef DEBUG
+     {
+       CFA_TO_NAME(a_cfa);
+       fprintf(stderr,"%08lx: does %08lx %.*s\n",
+               (Cell)ip,(Cell)a_cfa,len,name);
+     }
+ #endif
+ SET_IP(DOES_CODE1(a_cfa));
+ #endif
- branch-lp+!#    --      gforth  branch_lp_plus_store_number
+ \+glocals
+ branch-lp+!# ( #a_target #nlocals -- )  gforth  branch_lp_plus_store_number
  /* this will probably not be used */
- branch_adjust_lp:
+ lp += nlocals;
- lp += (Cell)(IP[1]);
+ #ifdef NO_IP
- goto branch;
+ INST_TAIL;
+ JUMP(a_target);
+ #else
+ SET_IP((Xt *)a_target);
+ #endif
- \+[THEN]
+ \+
- branch  --              gforth
+ branch  ( #a_target -- )        gforth
- branch:
+ #ifdef NO_IP
- ip = (Xt *)(((Cell)IP)+(Cell)NEXT_INST);
+ INST_TAIL;
- NEXT_P0;
+ JUMP(a_target);
+ #else
+ SET_IP((Xt *)a_target);
+ #endif
  :
-  r> dup @ + >r ;
+  r> @ >r ;
- \ condbranch(forthname,restline,code,forthcode)
+ \ condbranch(forthname,stackeffect,restline,code1,code2,forthcode)
+ \ this is non-syntactical: code must open a brace that is closed by the macro
+ \ condbranch(forthname,stackeffect,restline,code1,code2,forthcode)
  \ this is non-syntactical: code must open a brace that is closed by the macro
  define(condbranch,
- $1      $2
+ $1 ( `#'a_target $2 ) $3
- $3      ip = (Xt *)(((Cell)IP)+(Cell)NEXT_INST);
+ $4      #ifdef NO_IP
-         NEXT_P0;
+ INST_TAIL;
-         NEXT;
+ #endif
+ $5      #ifdef NO_IP
+ JUMP(a_target);
+ #else
+ SET_IP((Xt *)a_target);
+ #endif
  }
- else
+ $6
-     INC_IP(1);
- $4
- \+has? glocals [IF]
+ \+glocals
- $1-lp+!#        $2_lp_plus_store_number
+ $1-lp+!`#' ( `#'a_target `#'nlocals $2 ) $3_lp_plus_store_number
- $3    goto branch_adjust_lp;
+ $4      #ifdef NO_IP
+ INST_TAIL;
+ #endif
+ $5      lp += nlocals;
+ #ifdef NO_IP
+ JUMP(a_target);
+ #else
+ SET_IP((Xt *)a_target);
+ #endif
  }
- else
-     INC_IP(2);
- \+[THEN]
+ \+
  )
- condbranch(?branch,f --         f83     question_branch,
+ \ version that generates two jumps (not good for PR 15242 workaround)
- if (f==0) {
+ define(condbranch_twojump,
-     IF_TOS(TOS = sp[0]);
+ $1 ( `#'a_target $2 ) $3
+ $4      #ifdef NO_IP
+ INST_TAIL;
+ #endif
+ $5      #ifdef NO_IP
+ JUMP(a_target);
+ #else
+ SET_IP((Xt *)a_target);
+ INST_TAIL; NEXT_P2;
+ #endif
+ }
+ SUPER_CONTINUE;
+ $6
+ \+glocals
+ $1-lp+!`#' ( `#'a_target `#'nlocals $2 ) $3_lp_plus_store_number
+ $4      #ifdef NO_IP
+ INST_TAIL;
+ #endif
+ $5      lp += nlocals;
+ #ifdef NO_IP
+ JUMP(a_target);
+ #else
+ SET_IP((Xt *)a_target);
+ INST_TAIL; NEXT_P2;
+ #endif
+ }
+ SUPER_CONTINUE;
+ \+
+ )
+ condbranch(?branch,f --,f83     question_branch,
+ ,if (f==0) {
  ,:
-= dup     \ !f !f
+= dup 0=          \ !f f
-  r> dup @   \ !f !f IP branchoffset
+  r> tuck cell+      \ !f branchoffset f IP+
-  rot and +  \ !f IP|IP+branchoffset
+  and -rot @ and or  \ f&IP+|!f&branch
-  swap 0= cell and + \ IP''
   >r ;)
  \ we don't need an lp_plus_store version of the ?dup-stuff, because it
  \ is only used in if's (yet)
- \+has? xconds [IF]
+ \+xconds
- ?dup-?branch    f -- f  new     question_dupe_question_branch
+ ?dup-?branch    ( #a_target f -- S:... )        new     question_dupe_question_branch
  ""The run-time procedure compiled by @code{?DUP-IF}.""
  if (f==0) {
-   sp++;
+ #ifdef NO_IP
-   IF_TOS(TOS = sp[0]);
+ INST_TAIL;
-   ip = (Xt *)(((Cell)IP)+(Cell)NEXT_INST);
+ JUMP(a_target);
-   NEXT_P0;
+ #else
-   NEXT;
+ SET_IP((Xt *)a_target);
+ #endif
+ } else {
+ sp--;
+ sp[0]=f;
  }
- else
-   INC_IP(1);
- ?dup-0=-?branch f --    new     question_dupe_zero_equals_question_branch
+ ?dup-0=-?branch ( #a_target f -- S:... ) new    question_dupe_zero_equals_question_branch
  ""The run-time procedure compiled by @code{?DUP-0=-IF}.""
- /* the approach taken here of declaring the word as having the stack
- effect ( f -- ) and correcting for it in the branch-taken case costs a
- few cycles in that case, but is easy to convert to a CONDBRANCH
- invocation */
  if (f!=0) {
    sp--;
-   ip = (Xt *)(((Cell)IP)+(Cell)NEXT_INST);
+   sp[0]=f;
-   NEXT_P0;
+ #ifdef NO_IP
-   NEXT;
+   JUMP(a_target);
+ #else
+   SET_IP((Xt *)a_target);
+ #endif
  }
- else
-   INC_IP(1);
- \+[THEN]
+ \+
+ \fhas? skiploopprims 0= [IF]
- condbranch((next),--            cmFORTH paren_next,
+ condbranch((next),R:n1 -- R:n2,cmFORTH  paren_next,
- if ((*rp)--) {
+ n2=n1-1;
+ ,if (n1) {
  ,:
   r> r> dup 1- >r
-  IF dup @ + >r ELSE cell+ >r THEN ;)
+  IF @ >r ELSE cell+ >r THEN ;)
- condbranch((loop),--            gforth  paren_loop,
+ condbranch((loop),R:nlimit R:n1 -- R:nlimit R:n2,gforth paren_loop,
- Cell index = *rp+1;
+ n2=n1+1;
- Cell limit = rp[1];
+ ,if (n2 != nlimit) {
- if (index != limit) {
-     *rp = index;
  ,:
   r> r> 1+ r> 2dup =
   IF >r 1- >r cell+ >r
-  ELSE >r >r dup @ + >r THEN ;)
+  ELSE >r >r @ >r THEN ;)
- condbranch((+loop),n --         gforth  paren_plus_loop,
+ condbranch((+loop),n R:nlimit R:n1 -- R:nlimit R:n2,gforth paren_plus_loop,
  /* !! check this thoroughly */
- Cell index = *rp;
  /* sign bit manipulation and test: (x^y)<0 is equivalent to (x<0) != (y<0) */
  /* dependent upon two's complement arithmetic */
- Cell olddiff = index-rp[1];
+ Cell olddiff = n1-nlimit;
- if ((olddiff^(olddiff+n))>=0   /* the limit is not crossed */
+ n2=n1+n;
-     || (olddiff^n)>=0          /* it is a wrap-around effect */) {
+ ,if (((olddiff^(olddiff+n))    /* the limit is not crossed */
- #ifdef i386
+      &(olddiff^n))             /* OR it is a wrap-around effect */
-     *rp += n;
+     >=0) { /* & is used to avoid having two branches for gforth-native */
- #else
-     *rp = index + n;
- #endif
-     IF_TOS(TOS = sp[0]);
  ,:
   r> swap
   r> r> 2dup - >r
 pick r@ + r@ xor 0< 0=
 pick r> xor 0< 0= or
-  IF    >r + >r dup @ + >r
+  IF    >r + >r @ >r
   ELSE  >r >r drop cell+ >r THEN ;)
- \+has? xconds [IF]
+ \+xconds
- condbranch((-loop),u --         gforth  paren_minus_loop,
+ condbranch((-loop),u R:nlimit R:n1 -- R:nlimit R:n2,gforth paren_minus_loop,
- /* !! check this thoroughly */
+ UCell olddiff = n1-nlimit;
- Cell index = *rp;
+ n2=n1-u;
- UCell olddiff = index-rp[1];
+ ,if (olddiff>u) {
- if (olddiff>u) {
- #ifdef i386
-     *rp -= u;
- #else
-     *rp = index - u;
- #endif
-     IF_TOS(TOS = sp[0]);
  ,)
- condbranch((s+loop),n --                gforth  paren_symmetric_plus_loop,
+ condbranch((s+loop),n R:nlimit R:n1 -- R:nlimit R:n2,gforth     paren_symmetric_plus_loop,
  ""The run-time procedure compiled by S+LOOP. It loops until the index
  crosses the boundary between limit and limit-sign(n). I.e. a symmetric
  version of (+LOOP).""
  /* !! check this thoroughly */
- Cell index = *rp;
+ Cell diff = n1-nlimit;
- Cell diff = index-rp[1];
  Cell newdiff = diff+n;
  if (n<0) {
      diff = -diff;
      newdiff = -newdiff;
  }
- if (diff>=0 || newdiff<0) {
+ n2=n1+n;
- #ifdef i386
+ ,if (((~diff)|newdiff)<0) { /* use | to avoid two branches for gforth-native */
-     *rp += n;
- #else
-     *rp = index + n;
- #endif
-     IF_TOS(TOS = sp[0]);
  ,)
- \+[THEN]
+ \+
- unloop          --      core
+ (for)   ( ncount -- R:nlimit R:ncount )         cmFORTH         paren_for
- rp += 2;
- :
-  r> rdrop rdrop >r ;
- (for)   ncount --               cmFORTH         paren_for
  /* or (for) = >r -- collides with unloop! */
- *--rp = 0;
+ nlimit=0;
- *--rp = ncount;
  :
   r> swap 0 >r >r >r ;
- (do)    nlimit nstart --                gforth          paren_do
+ (do)    ( nlimit nstart -- R:nlimit R:nstart )  gforth          paren_do
- /* or do it in high-level? 0.09/0.23% */
- *--rp = nlimit;
- *--rp = nstart;
  :
   r> swap rot >r >r >r ;
- (?do)   nlimit nstart --        gforth  paren_question_do
+ (?do) ( #a_target nlimit nstart -- R:nlimit R:nstart ) gforth   paren_question_do
- *--rp = nlimit;
+ #ifdef NO_IP
- *--rp = nstart;
+     INST_TAIL;
+ #endif
  if (nstart == nlimit) {
-     IF_TOS(TOS = sp[0]);
+ #ifdef NO_IP
-     goto branch;
+     JUMP(a_target);
-     }
+ #else
- else {
+     SET_IP((Xt *)a_target);
-     INC_IP(1);
+ #endif
  }
  :
 dup =
    IF   r> swap rot >r >r
-        dup @ + >r
+        @ >r
    ELSE r> swap rot >r >r
         cell+ >r
    THEN ;                                \ --> CORE-EXT
- \+has? xconds [IF]
+ \+xconds
- (+do)   nlimit nstart --        gforth  paren_plus_do
+ (+do)   ( #a_target nlimit nstart -- R:nlimit R:nstart ) gforth paren_plus_do
- *--rp = nlimit;
+ #ifdef NO_IP
- *--rp = nstart;
+     INST_TAIL;
+ #endif
  if (nstart >= nlimit) {
-     IF_TOS(TOS = sp[0]);
+ #ifdef NO_IP
-     goto branch;
+     JUMP(a_target);
-     }
+ #else
- else {
+     SET_IP((Xt *)a_target);
-     INC_IP(1);
+ #endif
  }
  :
   swap 2dup
   r> swap >r swap >r
   >=
   IF
-      dup @ +
+      @
   ELSE
       cell+
   THEN  >r ;
- (u+do)  ulimit ustart --        gforth  paren_u_plus_do
+ (u+do)  ( #a_target ulimit ustart -- R:ulimit R:ustart ) gforth paren_u_plus_do
- *--rp = ulimit;
+ #ifdef NO_IP
- *--rp = ustart;
+     INST_TAIL;
+ #endif
  if (ustart >= ulimit) {
-     IF_TOS(TOS = sp[0]);
+ #ifdef NO_IP
-     goto branch;
+ JUMP(a_target);
-     }
+ #else
- else {
+ SET_IP((Xt *)a_target);
-     INC_IP(1);
+ #endif
  }
  :
   swap 2dup
   r> swap >r swap >r
   u>=
   IF
-      dup @ +
+      @
   ELSE
       cell+
   THEN  >r ;
- (-do)   nlimit nstart --        gforth  paren_minus_do
+ (-do)   ( #a_target nlimit nstart -- R:nlimit R:nstart ) gforth paren_minus_do
- *--rp = nlimit;
+ #ifdef NO_IP
- *--rp = nstart;
+     INST_TAIL;
+ #endif
  if (nstart <= nlimit) {
-     IF_TOS(TOS = sp[0]);
+ #ifdef NO_IP
-     goto branch;
+ JUMP(a_target);
-     }
+ #else
- else {
+ SET_IP((Xt *)a_target);
-     INC_IP(1);
+ #endif
  }
  :
   swap 2dup
   r> swap >r swap >r
   <=
   IF
-      dup @ +
+      @
   ELSE
       cell+
   THEN  >r ;
- (u-do)  ulimit ustart --        gforth  paren_u_minus_do
+ (u-do)  ( #a_target ulimit ustart -- R:ulimit R:ustart ) gforth paren_u_minus_do
- *--rp = ulimit;
+ #ifdef NO_IP
- *--rp = ustart;
+     INST_TAIL;
+ #endif
  if (ustart <= ulimit) {
-     IF_TOS(TOS = sp[0]);
+ #ifdef NO_IP
-     goto branch;
+ JUMP(a_target);
-     }
+ #else
- else {
+ SET_IP((Xt *)a_target);
-     INC_IP(1);
+ #endif
  }
  :
   swap 2dup
   r> swap >r swap >r
   u<=
   IF
-      dup @ +
+      @
   ELSE
       cell+
   THEN  >r ;
- \+[THEN]
+ \+
  \ don't make any assumptions where the return stack is!!
  \ implement this in machine code if it should run quickly!
- i       -- n            core
+ i       ( R:n -- R:n n )                core
- n = *rp;
  :
  \ rp@ cell+ @ ;
    r> r> tuck >r >r ;
- i'      -- w            gforth          i_tick
+ i'      ( R:w R:w2 -- R:w R:w2 w )              gforth          i_tick
- ""loop end value""
- w = rp[1];
  :
  \ rp@ cell+ cell+ @ ;
    r> r> r> dup itmp ! >r >r >r itmp @ ;
  variable itmp
- j       -- n            core
+ j       ( R:n R:d1 -- n R:n R:d1 )              core
- n = rp[2];
  :
  \ rp@ cell+ cell+ cell+ @ ;
    r> r> r> r> dup itmp ! >r >r >r >r itmp @ ;
  [IFUNDEF] itmp variable itmp [THEN]
- k       -- n            gforth
+ k       ( R:n R:d1 R:d2 -- n R:n R:d1 R:d2 )            gforth
- n = rp[4];
  :
  \ rp@ [ 5 cells ] Literal + @ ;
    r> r> r> r> r> r> dup itmp ! >r >r >r >r >r >r itmp @ ;
  [IFUNDEF] itmp variable itmp [THEN]
+ \f[THEN]
  \ digit is high-level: 0/0%
- move    c_from c_to ucount --           core
+ \g strings
+ move    ( c_from c_to ucount -- )               core
+ ""Copy the contents of @i{ucount} aus at @i{c-from} to
+ @i{c-to}. @code{move} works correctly even if the two areas overlap.""
+ /* !! note that the standard specifies addr, not c-addr */
  memmove(c_to,c_from,ucount);
  /* make an Ifdef for bsd and others? */
  :
   >r 2dup u< IF r> cmove> ELSE r> cmove THEN ;
- cmove   c_from c_to u --        string
+ cmove   ( c_from c_to u -- )    string  c_move
- while (u-- > 0)
+ ""Copy the contents of @i{ucount} characters from data space at
-   *c_to++ = *c_from++;
+ @i{c-from} to @i{c-to}. The copy proceeds @code{char}-by-@code{char}
+ from low address to high address; i.e., for overlapping areas it is
+ safe if @i{c-to}=<@i{c-from}.""
+ cmove(c_from,c_to,u);
  :
   bounds ?DO  dup c@ I c! 1+  LOOP  drop ;
- cmove>  c_from c_to u --        string  c_move_up
+ cmove>  ( c_from c_to u -- )    string  c_move_up
- while (u-- > 0)
+ ""Copy the contents of @i{ucount} characters from data space at
-   c_to[u] = c_from[u];
+ @i{c-from} to @i{c-to}. The copy proceeds @code{char}-by-@code{char}
+ from high address to low address; i.e., for overlapping areas it is
+ safe if @i{c-to}>=@i{c-from}.""
+ cmove_up(c_from,c_to,u);
  :
   dup 0= IF  drop 2drop exit  THEN
   rot over + -rot bounds swap 1-
   DO  1- dup c@ I c!  -1 +LOOP  drop ;
- fill    c_addr u c --   core
+ fill    ( c_addr u c -- )       core
+ ""Store @i{c} in @i{u} chars starting at @i{c-addr}.""
  memset(c_addr,c,u);
  :
   -rot bounds
   ?DO  dup I c!  LOOP  drop ;
- compare         c_addr1 u1 c_addr2 u2 -- n      string
+ compare ( c_addr1 u1 c_addr2 u2 -- n )  string
- ""Compare the strings lexicographically. If they are equal, n is 0; if
+ ""Compare two strings lexicographically. If they are equal, @i{n} is 0; if
- the first string is smaller, n is -1; if the first string is larger, n
+ the first string is smaller, @i{n} is -1; if the first string is larger, @i{n}
  is 1. Currently this is based on the machine's character
- comparison. In the future, this may change to considering the current
+ comparison. In the future, this may change to consider the current
  locale and its collation order.""
- n = memcmp(c_addr1, c_addr2, u1<u2 ? u1 : u2);
+ /* close ' to keep fontify happy */
- if (n==0)
+ n = compare(c_addr1, u1, c_addr2, u2);
-   n = u1-u2;
- if (n<0)
-   n = -1;
- else if (n>0)
-   n = 1;
- :
-  rot 2dup - >r min swap -text dup
-  IF    rdrop
-  ELSE  drop r@ 0>
-        IF    rdrop -1
-        ELSE  r> 1 and
-        THEN
-  THEN ;
- -text           c_addr1 u c_addr2 -- n  new     dash_text
- n = memcmp(c_addr1, c_addr2, u);
- if (n<0)
-   n = -1;
- else if (n>0)
-   n = 1;
  :
+  rot 2dup swap - >r min swap -text dup
+  IF  rdrop  ELSE  drop r> sgn  THEN ;
+ : -text ( c_addr1 u c_addr2 -- n )
   swap bounds
   ?DO  dup c@ I c@ = WHILE  1+  LOOP  drop 0
-  ELSE  c@ I c@ - unloop  THEN  -text-flag ;
+  ELSE  c@ I c@ - unloop  THEN  sgn ;
- : -text-flag ( n -- -1/0/1 )
+ : sgn ( n -- -1/0/1 )
-  dup 0< IF  drop -1  ELSE  0>  1 and  THEN  ;
+  dup 0= IF EXIT THEN  0< 2* 1+ ;
- toupper c1 -- c2        gforth
+ \ -text is only used by replaced primitives now; move it elsewhere
+ \ -text ( c_addr1 u c_addr2 -- n )      new     dash_text
+ \ n = memcmp(c_addr1, c_addr2, u);
+ \ if (n<0)
+ \   n = -1;
+ \ else if (n>0)
+ \   n = 1;
+ \ :
+ \  swap bounds
+ \  ?DO  dup c@ I c@ = WHILE  1+  LOOP  drop 0
+ \  ELSE  c@ I c@ - unloop  THEN  sgn ;
+ \ : sgn ( n -- -1/0/1 )
+ \  dup 0= IF EXIT THEN  0< 2* 1+ ;
+ toupper ( c1 -- c2 )    gforth
+ ""If @i{c1} is a lower-case character (in the current locale), @i{c2}
+ is the equivalent upper-case character. All other characters are unchanged.""
  c2 = toupper(c1);
  :
   dup [char] a - [ char z char a - 1 + ] Literal u<  bl and - ;
- capscomp        c_addr1 u c_addr2 -- n  new
+ capscompare     ( c_addr1 u1 c_addr2 u2 -- n )  string
- n = memcasecmp(c_addr1, c_addr2, u); /* !! use something that works in all locales */
+ ""Compare two strings lexicographically. If they are equal, @i{n} is 0; if
- if (n<0)
+ the first string is smaller, @i{n} is -1; if the first string is larger, @i{n}
-   n = -1;
+ is 1. Currently this is based on the machine's character
- else if (n>0)
+ comparison. In the future, this may change to consider the current
-   n = 1;
+ locale and its collation order.""
- :
+ /* close ' to keep fontify happy */
-  swap bounds
+ n = capscompare(c_addr1, u1, c_addr2, u2);
-  ?DO  dup c@ I c@ <>
-      IF  dup c@ toupper I c@ toupper =
-      ELSE  true  THEN  WHILE  1+  LOOP  drop 0
-  ELSE  c@ toupper I c@ toupper - unloop  THEN  -text-flag ;
- -trailing       c_addr u1 -- c_addr u2          string  dash_trailing
- u2 = u1;
- while (u2>0 && c_addr[u2-1] == ' ')
-   u2--;
- :
-  BEGIN  1- 2dup + c@ bl =  WHILE
-         dup  0= UNTIL  ELSE  1+  THEN ;
- /string         c_addr1 u1 n -- c_addr2 u2      string  slash_string
+ /string ( c_addr1 u1 n -- c_addr2 u2 )  string  slash_string
+ ""Adjust the string specified by @i{c-addr1, u1} to remove @i{n}
+ characters from the start of the string.""
  c_addr2 = c_addr1+n;
  u2 = u1-n;
  :
   tuck - >r + r> dup 0< IF  - 0  THEN ;
- +       n1 n2 -- n              core    plus
+ \g arith
+ lit     ( #w -- w )             gforth
+ :
+  r> dup @ swap cell+ >r ;
+ +       ( n1 n2 -- n )          core    plus
+ n = n1+n2;
+ \ lit+ / lit_plus = lit +
+ lit+    ( n1 #n2 -- n )         new     lit_plus
  n = n1+n2;
  \ PFE-0.9.14 has it differently, but the next release will have it as follows
- under+  n1 n2 n3 -- n n2        gforth  under_plus
+ under+  ( n1 n2 n3 -- n n2 )    gforth  under_plus
- ""add @var{n3} to @var{n1} (giving @var{n})""
+ ""add @i{n3} to @i{n1} (giving @i{n})""
  n = n1+n3;
  :
   rot + swap ;
- -       n1 n2 -- n              core    minus
+ -       ( n1 n2 -- n )          core    minus
  n = n1-n2;
  :
   negate + ;
- negate  n1 -- n2                core
+ negate  ( n1 -- n2 )            core
  /* use minus as alias */
  n2 = -n1;
  :
   invert 1+ ;
-+      n1 -- n2                core            one_plus
++      ( n1 -- n2 )            core            one_plus
  n2 = n1+1;
  :
 + ;
--      n1 -- n2                core            one_minus
+-      ( n1 -- n2 )            core            one_minus
  n2 = n1-1;
  :
 - ;
- max     n1 n2 -- n      core
+ max     ( n1 n2 -- n )  core
  if (n1<n2)
    n = n2;
  else
- Line 574
+ Line 786
  Line 574
  Line 786
  :
 dup < IF swap THEN drop ;
- min     n1 n2 -- n      core
+ min     ( n1 n2 -- n )  core
  if (n1<n2)
    n = n1;
  else
- Line 582
+ Line 794
  Line 582
  Line 794
  :
 dup > IF swap THEN drop ;
- abs     n1 -- n2        core
+ abs     ( n -- u )      core
- if (n1<0)
+ if (n<0)
-   n2 = -n1;
+   u = -n;
  else
-   n2 = n1;
+   u = n;
  :
   dup 0< IF negate THEN ;
- *       n1 n2 -- n              core    star
+ *       ( n1 n2 -- n )          core    star
  n = n1*n2;
  :
   um* drop ;
- /       n1 n2 -- n              core    slash
+ /       ( n1 n2 -- n )          core    slash
  n = n1/n2;
+ if (CHECK_DIVISION_SW && n2 == 0)
+   throw(BALL_DIVZERO);
+ if (CHECK_DIVISION_SW && n2 == -1 && n1 == CELL_MIN)
+   throw(BALL_RESULTRANGE);
+ if (FLOORED_DIV && ((n1^n2) < 0) && (n1%n2 != 0))
+   n--;
  :
   /mod nip ;
- mod     n1 n2 -- n              core
+ mod     ( n1 n2 -- n )          core
  n = n1%n2;
+ if (CHECK_DIVISION_SW && n2 == 0)
+   throw(BALL_DIVZERO);
+ if (CHECK_DIVISION_SW && n2 == -1 && n1 == CELL_MIN)
+   throw(BALL_RESULTRANGE);
+ if(FLOORED_DIV && ((n1^n2) < 0) && n!=0) n += n2;
  :
   /mod drop ;
- /mod    n1 n2 -- n3 n4          core            slash_mod
+ /mod    ( n1 n2 -- n3 n4 )              core            slash_mod
  n4 = n1/n2;
  n3 = n1%n2; /* !! is this correct? look into C standard! */
+ if (CHECK_DIVISION_SW && n2 == 0)
+   throw(BALL_DIVZERO);
+ if (CHECK_DIVISION_SW && n2 == -1 && n1 == CELL_MIN)
+   throw(BALL_RESULTRANGE);
+ if (FLOORED_DIV && ((n1^n2) < 0) && n3!=0) {
+   n4--;
+   n3+=n2;
+ }
  :
   >r s>d r> fm/mod ;
-*      n1 -- n2                core            two_star
+ */mod   ( n1 n2 n3 -- n4 n5 )   core    star_slash_mod
+ ""n1*n2=n3*n5+n4, with the intermediate result (n1*n2) being double.""
+ #ifdef BUGGY_LL_MUL
+ DCell d = mmul(n1,n2);
+ #else
+ DCell d = (DCell)n1 * (DCell)n2;
+ #endif
+ #ifdef BUGGY_LL_DIV
+ DCell r = fmdiv(d,n3);
+ n4=DHI(r);
+ n5=DLO(r);
+ #else
+ /* assumes that the processor uses either floored or symmetric division */
+ DCell d5 = d/n3;
+ n4 = d%n3;
+ if (CHECK_DIVISION_SW && n3 == 0)
+   throw(BALL_DIVZERO);
+ if (FLOORED_DIV && ((DHI(d)^n3)<0) && n4!=0) {
+   d5--;
+   n4+=n3;
+ }
+ n5 = d5;
+ if (CHECK_DIVISION && d5 != n5)
+   throw(BALL_RESULTRANGE);
+ #endif
+ :
+  >r m* r> fm/mod ;
+ */      ( n1 n2 n3 -- n4 )      core    star_slash
+ ""n4=(n1*n2)/n3, with the intermediate result being double.""
+ #ifdef BUGGY_LL_MUL
+ DCell d = mmul(n1,n2);
+ #else
+ DCell d = (DCell)n1 * (DCell)n2;
+ #endif
+ #ifdef BUGGY_LL_DIV
+ DCell r = fmdiv(d,n3);
+ n4=DLO(r);
+ #else
+ /* assumes that the processor uses either floored or symmetric division */
+ DCell d4 = d/n3;
+ if (CHECK_DIVISION_SW && n3 == 0)
+   throw(BALL_DIVZERO);
+ if (FLOORED_DIV && ((DHI(d)^n3)<0) && (d%n3)!=0)
+   d4--;
+ n4 = d4;
+ if (CHECK_DIVISION && d4 != n4)
+   throw(BALL_RESULTRANGE);
+ #endif
+ :
+  */mod nip ;
+*      ( n1 -- n2 )            core            two_star
+ ""Shift left by 1; also works on unsigned numbers""
  n2 = 2*n1;
  :
   dup + ;
-/      n1 -- n2                core            two_slash
+/      ( n1 -- n2 )            core            two_slash
- /* !! is this still correct? */
+ ""Arithmetic shift right by 1.  For signed numbers this is a floored
+ division by 2 (note that @code{/} not necessarily floors).""
  n2 = n1>>1;
  :
   dup MINI and IF 1 ELSE 0 THEN
-  [ bits/byte cell * 1- ] literal
+  [ bits/char cell * 1- ] literal
 DO 2* swap dup 2* >r MINI and
       IF 1 ELSE 0 THEN or r> swap
   LOOP nip ;
- fm/mod  d1 n1 -- n2 n3          core            f_m_slash_mod
+ fm/mod  ( d1 n1 -- n2 n3 )              core            f_m_slash_mod
- ""floored division: d1 = n3*n1+n2, n1>n2>=0 or 0>=n2>n1""
+ ""Floored division: @i{d1} = @i{n3}*@i{n1}+@i{n2}, @i{n1}>@i{n2}>=0 or 0>=@i{n2}>@i{n1}.""
- #ifdef BUGGY_LONG_LONG
+ #ifdef ASM_SM_SLASH_REM
- DCell r = fmdiv(d1,n1);
+ #ifdef BUGGY_LL_DIV
- n2=r.hi;
+ ASM_SM_SLASH_REM(d1.lo, d1.hi, n1, n2, n3);
- n3=r.lo;
+ if (((DHI(d1)^n1)<0) && n2!=0) {
+   if (CHECK_DIVISION && n3 == CELL_MIN)
+     throw(BALL_RESULTRANGE);
+   n3--;
+   n2+=n1;
+ }
  #else
- /* assumes that the processor uses either floored or symmetric division */
+ ASM_SM_SLASH_REM4(d1, n1, n2, n3);
- n3 = d1/n1;
+ if (((DHI(d1)^n1)<0) && n2!=0) {
- n2 = d1%n1;
+   if (CHECK_DIVISION && n3 == CELL_MIN)
- /* note that this 1%-3>0 is optimized by the compiler */
+     throw(BALL_RESULTRANGE);
- if (1%-3>0 && (d1<0) != (n1<0) && n2!=0) {
    n3--;
    n2+=n1;
  }
  #endif
+ #else /* !defined(ASM_SM_SLASH_REM) */
+ DCell r = fmdiv(d1,n1);
+ n2=DHI(r);
+ n3=DLO(r);
+ #endif /* !defined(ADM_SM_SLASH_REM) */
  :
   dup >r dup 0< IF  negate >r dnegate r>  THEN
   over       0< IF  tuck + swap  THEN
   um/mod
   r> 0< IF  swap negate swap  THEN ;
- sm/rem  d1 n1 -- n2 n3          core            s_m_slash_rem
+ sm/rem  ( d1 n1 -- n2 n3 )              core            s_m_slash_rem
- ""symmetric division: d1 = n3*n1+n2, sign(n2)=sign(d1) or 0""
+ ""Symmetric division: @i{d1} = @i{n3}*@i{n1}+@i{n2}, sign(@i{n2})=sign(@i{d1}) or 0.""
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_DIV
+ #ifdef ASM_SM_SLASH_REM
+ ASM_SM_SLASH_REM(d1.lo, d1.hi, n1, n2, n3);
+ #else /* !defined(ASM_SM_SLASH_REM) */
  DCell r = smdiv(d1,n1);
- n2=r.hi;
+ n2=DHI(r);
- n3=r.lo;
+ n3=DLO(r);
- #else
+ #endif /* !defined(ASM_SM_SLASH_REM) */
+ #else
+ #ifdef ASM_SM_SLASH_REM4
+ ASM_SM_SLASH_REM4(d1, n1, n2, n3);
+ #else /* !defined(ASM_SM_SLASH_REM4) */
  /* assumes that the processor uses either floored or symmetric division */
- n3 = d1/n1;
+ DCell d3 = d1/n1;
  n2 = d1%n1;
+ if (CHECK_DIVISION_SW && n1 == 0)
+   throw(BALL_DIVZERO);
  /* note that this 1%-3<0 is optimized by the compiler */
- if (1%-3<0 && (d1<0) != (n1<0) && n2!=0) {
+ if (1%-3<0 && ((DHI(d1)^n1)<0) && n2!=0) {
-   n3++;
+   d3++;
    n2-=n1;
  }
+ n3 = d3;
+ if (CHECK_DIVISION && d3 != n3)
+   throw(BALL_RESULTRANGE);
+ #endif /* !defined(ASM_SM_SLASH_REM4) */
  #endif
  :
   over >r dup >r abs -rot
- Line 670
+ Line 977
  Line 670
  Line 977
   r> r@ xor 0< IF       negate       THEN
   r>        0< IF  swap negate swap  THEN ;
- m*      n1 n2 -- d              core    m_star
+ m*      ( n1 n2 -- d )          core    m_star
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_MUL
  d = mmul(n1,n2);
  #else
  d = (DCell)n1 * (DCell)n2;
- Line 681
+ Line 988
  Line 681
  Line 988
 dup swap 0< and >r
   um* r> - r> - ;
- um*     u1 u2 -- ud             core    u_m_star
+ um*     ( u1 u2 -- ud )         core    u_m_star
  /* use u* as alias */
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_MUL
  ud = ummul(u1,u2);
  #else
  ud = (UDCell)u1 * (UDCell)u2;
  #endif
  :
-    >r >r 0 0 r> r> [ 8 cells ] literal 0
+-rot dup [ 8 cells ] literal -
     DO
-        over >r dup >r 0< and d2*+ drop
+         dup 0< I' and d2*+ drop
-        r> 2* r> swap
+    LOOP ;
-    LOOP 2drop ;
  : d2*+ ( ud n -- ud+n c )
     over MINI
     and >r >r 2dup d+ swap r> + swap r> ;
- um/mod  ud u1 -- u2 u3          core    u_m_slash_mod
+ um/mod  ( ud u1 -- u2 u3 )              core    u_m_slash_mod
- #ifdef BUGGY_LONG_LONG
+ ""ud=u3*u1+u2, u1>u2>=0""
+ #ifdef BUGGY_LL_DIV
+ #ifdef ASM_UM_SLASH_MOD
+ ASM_UM_SLASH_MOD(ud.lo, ud.hi, u1, u2, u3);
+ #else /* !defined(ASM_UM_SLASH_MOD) */
  UDCell r = umdiv(ud,u1);
- u2=r.hi;
+ u2=DHI(r);
- u3=r.lo;
+ u3=DLO(r);
- #else
+ #endif /* !defined(ASM_UM_SLASH_MOD) */
- u3 = ud/u1;
+ #else
+ #ifdef ASM_UM_SLASH_MOD4
+ ASM_UM_SLASH_MOD4(ud, u1, u2, u3);
+ #else /* !defined(ASM_UM_SLASH_MOD4) */
+ UDCell ud3 = ud/u1;
  u2 = ud%u1;
+ if (CHECK_DIVISION_SW && u1 == 0)
+   throw(BALL_DIVZERO);
+ u3 = ud3;
+ if (CHECK_DIVISION && ud3 != u3)
+   throw(BALL_RESULTRANGE);
+ #endif /* !defined(ASM_UM_SLASH_MOD4) */
  #endif
  :
 swap [ 8 cells 1 + ] literal 0
- Line 717
+ Line 1037
  Line 717
  Line 1037
     over MINI
     and >r >r 2dup d+ swap r> + swap r> ;
- m+      d1 n -- d2              double          m_plus
+ m+      ( d1 n -- d2 )          double          m_plus
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_ADD
- d2.lo = d1.lo+n;
+ DLO_IS(d2, DLO(d1)+n);
- d2.hi = d1.hi - (n<0) + (d2.lo<d1.lo);
+ DHI_IS(d2, DHI(d1) - (n<0) + (DLO(d2)<DLO(d1)));
  #else
  d2 = d1+n;
  #endif
  :
   s>d d+ ;
- d+      d1 d2 -- d              double  d_plus
+ d+      ( d1 d2 -- d )          double  d_plus
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_ADD
- d.lo = d1.lo+d2.lo;
+ DLO_IS(d, DLO(d1) + DLO(d2));
- d.hi = d1.hi + d2.hi + (d.lo<d1.lo);
+ DHI_IS(d, DHI(d1) + DHI(d2) + (d.lo<DLO(d1)));
  #else
  d = d1+d2;
  #endif
  :
   rot + >r tuck + swap over u> r> swap - ;
- d-      d1 d2 -- d              double          d_minus
+ d-      ( d1 d2 -- d )          double          d_minus
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_ADD
- d.lo = d1.lo - d2.lo;
+ DLO_IS(d, DLO(d1) - DLO(d2));
- d.hi = d1.hi-d2.hi-(d1.lo<d2.lo);
+ DHI_IS(d, DHI(d1)-DHI(d2)-(DLO(d1)<DLO(d2)));
  #else
  d = d1-d2;
  #endif
  :
   dnegate d+ ;
- dnegate d1 -- d2                double
+ dnegate ( d1 -- d2 )            double  d_negate
  /* use dminus as alias */
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_ADD
  d2 = dnegate(d1);
  #else
  d2 = -d1;
- Line 757
+ Line 1077
  Line 757
  Line 1077
  :
   invert swap negate tuck 0= - ;
- d2*     d1 -- d2                double          d_two_star
+ d2*     ( d1 -- d2 )            double          d_two_star
- #ifdef BUGGY_LONG_LONG
+ ""Shift left by 1; also works on unsigned numbers""
- d2.lo = d1.lo<<1;
+ #ifdef BUGGY_LL_SHIFT
- d2.hi = (d1.hi<<1) | (d1.lo>>(CELL_BITS-1));
+ DLO_IS(d2, DLO(d1)<<1);
+ DHI_IS(d2, (DHI(d1)<<1) | (DLO(d1)>>(CELL_BITS-1)));
  #else
  d2 = 2*d1;
  #endif
  :
 dup d+ ;
- d2/     d1 -- d2                double          d_two_slash
+ d2/     ( d1 -- d2 )            double          d_two_slash
- #ifdef BUGGY_LONG_LONG
+ ""Arithmetic shift right by 1.  For signed numbers this is a floored
- d2.hi = d1.hi>>1;
+ division by 2.""
- d2.lo= (d1.lo>>1) | (d1.hi<<(CELL_BITS-1));
+ #ifdef BUGGY_LL_SHIFT
+ DHI_IS(d2, DHI(d1)>>1);
+ DLO_IS(d2, (DLO(d1)>>1) | (DHI(d1)<<(CELL_BITS-1)));
  #else
  d2 = d1>>1;
  #endif
- Line 778
+ Line 1101
  Line 778
  Line 1101
   dup 1 and >r 2/ swap 2/ [ 1 8 cells 1- lshift 1- ] Literal and
   r> IF  [ 1 8 cells 1- lshift ] Literal + THEN  swap ;
- and     w1 w2 -- w              core
+ and     ( w1 w2 -- w )          core
  w = w1&w2;
- or      w1 w2 -- w              core
+ or      ( w1 w2 -- w )          core
  w = w1|w2;
  :
   invert swap invert and invert ;
- xor     w1 w2 -- w              core
+ xor     ( w1 w2 -- w )          core    x_or
  w = w1^w2;
- invert  w1 -- w2                core
+ invert  ( w1 -- w2 )            core
  w2 = ~w1;
  :
   MAXU xor ;
- rshift  u1 n -- u2              core
+ rshift  ( u1 n -- u2 )          core    r_shift
+ ""Logical shift right by @i{n} bits.""
+ #ifdef BROKEN_SHIFT
+   u2 = rshift(u1, n);
+ #else
    u2 = u1>>n;
+ #endif
  :
 ?DO 2/ MAXI and LOOP ;
- lshift  u1 n -- u2              core
+ lshift  ( u1 n -- u2 )          core    l_shift
+ #ifdef BROKEN_SHIFT
+   u2 = lshift(u1, n);
+ #else
    u2 = u1<<n;
+ #endif
  :
 ?DO 2* LOOP ;
+ \g compare
  \ comparisons(prefix, args, prefix, arg1, arg2, wordsets...)
  define(comparisons,
- $1=     $2 -- f         $6      $3equals
+ $1=     ( $2 -- f )             $6      $3equals
  f = FLAG($4==$5);
  :
      [ char $1x char 0 = [IF]
- Line 815
+ Line 1149
  Line 815
  Line 1149
          ] xor 0= [
      [THEN] ] ;
- $1<>    $2 -- f         $7      $3different
+ $1<>    ( $2 -- f )             $7      $3not_equals
  f = FLAG($4!=$5);
  :
      [ char $1x char 0 = [IF]
- Line 824
+ Line 1158
  Line 824
  Line 1158
          ] xor 0<> [
      [THEN] ] ;
- $1<     $2 -- f         $8      $3less
+ $1<     ( $2 -- f )             $8      $3less_than
  f = FLAG($4<$5);
  :
      [ char $1x char 0 = [IF]
- Line 836
+ Line 1170
  Line 836
  Line 1170
          [THEN]
      [THEN] ] ;
- $1>     $2 -- f         $9      $3greater
+ $1>     ( $2 -- f )             $9      $3greater_than
  f = FLAG($4>$5);
  :
      [ char $1x char 0 = [IF] ] negate [ [ELSE] ] swap [ [THEN] ]
      $1< ;
- $1<=    $2 -- f         gforth  $3less_or_equal
+ $1<=    ( $2 -- f )             gforth  $3less_or_equal
  f = FLAG($4<=$5);
  :
      $1> 0= ;
- $1>=    $2 -- f         gforth  $3greater_or_equal
+ $1>=    ( $2 -- f )             gforth  $3greater_or_equal
  f = FLAG($4>=$5);
  :
      [ char $1x char 0 = [IF] ] negate [ [ELSE] ] swap [ [THEN] ]
- Line 861
+ Line 1195
  Line 861
  Line 1195
  \ dcomparisons(prefix, args, prefix, arg1, arg2, wordsets...)
  define(dcomparisons,
- $1=     $2 -- f         $6      $3equals
+ $1=     ( $2 -- f )             $6      $3equals
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_CMP
  f = FLAG($4.lo==$5.lo && $4.hi==$5.hi);
  #else
  f = FLAG($4==$5);
  #endif
- $1<>    $2 -- f         $7      $3different
+ $1<>    ( $2 -- f )             $7      $3not_equals
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_CMP
  f = FLAG($4.lo!=$5.lo || $4.hi!=$5.hi);
  #else
  f = FLAG($4!=$5);
  #endif
- $1<     $2 -- f         $8      $3less
+ $1<     ( $2 -- f )             $8      $3less_than
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_CMP
  f = FLAG($4.hi==$5.hi ? $4.lo<$5.lo : $4.hi<$5.hi);
  #else
  f = FLAG($4<$5);
  #endif
- $1>     $2 -- f         $9      $3greater
+ $1>     ( $2 -- f )             $9      $3greater_than
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_CMP
  f = FLAG($4.hi==$5.hi ? $4.lo>$5.lo : $4.hi>$5.hi);
  #else
  f = FLAG($4>$5);
  #endif
- $1<=    $2 -- f         gforth  $3less_or_equal
+ $1<=    ( $2 -- f )             gforth  $3less_or_equal
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_CMP
  f = FLAG($4.hi==$5.hi ? $4.lo<=$5.lo : $4.hi<=$5.hi);
  #else
  f = FLAG($4<=$5);
  #endif
- $1>=    $2 -- f         gforth  $3greater_or_equal
+ $1>=    ( $2 -- f )             gforth  $3greater_or_equal
- #ifdef BUGGY_LONG_LONG
+ #ifdef BUGGY_LL_CMP
  f = FLAG($4.hi==$5.hi ? $4.lo>=$5.lo : $4.hi>=$5.hi);
  #else
  f = FLAG($4>=$5);
- Line 905
+ Line 1239
  Line 905
  Line 1239
  )
- \+has? dcomps [IF]
+ \+dcomps
  dcomparisons(d, d1 d2, d_, d1, d2, double, gforth, double, gforth)
  dcomparisons(d0, d, d_zero_, d, DZERO, double, gforth, double, gforth)
  dcomparisons(du, ud1 ud2, d_u_, ud1, ud2, gforth, gforth, double-ext, gforth)
- \+[THEN]
+ \+
- within  u1 u2 u3 -- f           core-ext
+ within  ( u1 u2 u3 -- f )               core-ext
+ ""u2=<u1<u3 or: u3=<u2 and u1 is not in [u3,u2).  This works for
+ unsigned and signed numbers (but not a mixture).  Another way to think
+ about this word is to consider the numbers as a circle (wrapping
+ around from @code{max-u} to 0 for unsigned, and from @code{max-n} to
+ min-n for signed numbers); now consider the range from u2 towards
+ increasing numbers up to and excluding u3 (giving an empty range if
+ u2=u3); if u1 is in this range, @code{within} returns true.""
  f = FLAG(u1-u2 < u3-u2);
  :
   over - >r - r> u< ;
- sp@     -- a_addr               gforth          spat
+ \g stack
- a_addr = sp+1;
- sp!     a_addr --               gforth          spstore
+ useraddr        ( #u -- a_addr )        new
+ a_addr = (Cell *)(up+u);
+ up!     ( a_addr -- )   gforth  up_store
+ gforth_UP=up=(Address)a_addr;
+ :
+  up ! ;
+ Variable UP
+ sp@     ( S:... -- a_addr )             gforth          sp_fetch
+ a_addr = sp;
+ sp!     ( a_addr -- S:... )             gforth          sp_store
  sp = a_addr;
- /* works with and without TOS caching */
- rp@     -- a_addr               gforth          rpat
+ rp@     ( -- a_addr )           gforth          rp_fetch
  a_addr = rp;
- rp!     a_addr --               gforth          rpstore
+ rp!     ( a_addr -- )           gforth          rp_store
  rp = a_addr;
- \+has? floating [IF]
+ \+floating
- fp@     -- f_addr       gforth  fp_fetch
+ fp@     ( f:... -- f_addr )     gforth  fp_fetch
  f_addr = fp;
- fp!     f_addr --       gforth  fp_store
+ fp!     ( f_addr -- f:... )     gforth  fp_store
  fp = f_addr;
- \+[THEN]
+ \+
- ;s      --              gforth  semis
- ip = (Xt *)(*rp++);
- NEXT_P0;
- >r      w --            core    to_r
+ >r      ( w -- R:w )            core    to_r
- *--rp = w;
  :
   (>r) ;
  : (>r)  rp@ cell+ @ rp@ ! rp@ cell+ ! ;
- r>      -- w            core    r_from
+ r>      ( R:w -- w )            core    r_from
- w = *rp++;
  :
   rp@ cell+ @ rp@ @ rp@ cell+ ! (rdrop) rp@ ! ;
  Create (rdrop) ' ;s A,
- rdrop   --              gforth
+ rdrop   ( R:w -- )              gforth
- rp++;
  :
   r> r> drop >r ;
->r     w1 w2 --        core-ext        two_to_r
+>r     ( d -- R:d )    core-ext        two_to_r
- *--rp = w1;
- *--rp = w2;
  :
   swap r> swap >r swap >r >r ;
-r>     -- w1 w2        core-ext        two_r_from
+r>     ( R:d -- d )    core-ext        two_r_from
- w2 = *rp++;
- w1 = *rp++;
  :
   r> r> swap r> swap >r swap ;
-r@     -- w1 w2        core-ext        two_r_fetch
+r@     ( R:d -- R:d d )        core-ext        two_r_fetch
- w2 = rp[0];
- w1 = rp[1];
  :
   i' j ;
-rdrop  --              gforth  two_r_drop
+rdrop  ( R:d -- )              gforth  two_r_drop
- rp+=2;
  :
   r> r> drop r> drop >r ;
- over    w1 w2 -- w1 w2 w1               core
+ over    ( w1 w2 -- w1 w2 w1 )           core
  :
   sp@ cell+ @ ;
- drop    w --            core
+ drop    ( w -- )                core
  :
   IF THEN ;
- swap    w1 w2 -- w2 w1          core
+ swap    ( w1 w2 -- w2 w1 )              core
  :
   >r (swap) ! r> (swap) @ ;
  Variable (swap)
- dup     w -- w w                core
+ dup     ( w -- w w )            core    dupe
  :
   sp@ @ ;
- rot     w1 w2 w3 -- w2 w3 w1    core    rote
+ rot     ( w1 w2 w3 -- w2 w3 w1 )        core    rote
  :
  [ defined? (swap) [IF] ]
      (swap) ! (rot) ! >r (rot) @ (swap) @ r> ;
- Line 1011
+ Line 1348
  Line 1011
  Line 1348
      >r swap r> swap ;
  [THEN]
- -rot    w1 w2 w3 -- w3 w1 w2    gforth  not_rote
+ -rot    ( w1 w2 w3 -- w3 w1 w2 )        gforth  not_rote
  :
   rot rot ;
- nip     w1 w2 -- w2             core-ext
+ nip     ( w1 w2 -- w2 )         core-ext
  :
   swap drop ;
- tuck    w1 w2 -- w2 w1 w2       core-ext
+ tuck    ( w1 w2 -- w2 w1 w2 )   core-ext
  :
   swap over ;
- ?dup    w -- w                  core    question_dupe
+ ?dup    ( w -- S:... w )        core    question_dupe
+ ""Actually the stack effect is: @code{( w -- 0 | w w )}.  It performs a
+ @code{dup} if w is nonzero.""
  if (w!=0) {
-   IF_TOS(*sp-- = w;)
- #ifndef USE_TOS
    *--sp = w;
- #endif
  }
  :
   dup IF dup THEN ;
- pick    u -- w                  core-ext
+ pick    ( S:... u -- S:... w )          core-ext
- w = sp[u+1];
+ ""Actually the stack effect is @code{ x0 ... xu u -- x0 ... xu x0 }.""
+ w = sp[u];
  :
 + cells sp@ + @ ;
-drop   w1 w2 --                core    two_drop
+drop   ( w1 w2 -- )            core    two_drop
  :
   drop drop ;
-dup    w1 w2 -- w1 w2 w1 w2    core    two_dupe
+dup    ( w1 w2 -- w1 w2 w1 w2 )        core    two_dupe
  :
   over over ;
-over   w1 w2 w3 w4 -- w1 w2 w3 w4 w1 w2        core    two_over
+over   ( w1 w2 w3 w4 -- w1 w2 w3 w4 w1 w2 )    core    two_over
  :
 pick 3 pick ;
-swap   w1 w2 w3 w4 -- w3 w4 w1 w2      core    two_swap
+swap   ( w1 w2 w3 w4 -- w3 w4 w1 w2 )  core    two_swap
  :
   rot >r rot r> ;
-rot    w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2  double-ext      two_rote
+rot    ( w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2 )      double-ext      two_rote
  :
   >r >r 2swap r> r> 2swap ;
-nip    w1 w2 w3 w4 -- w3 w4    gforth  two_nip
+nip    ( w1 w2 w3 w4 -- w3 w4 )        gforth  two_nip
  :
 swap 2drop ;
-tuck   w1 w2 w3 w4 -- w3 w4 w1 w2 w3 w4        gforth  two_tuck
+tuck   ( w1 w2 w3 w4 -- w3 w4 w1 w2 w3 w4 )    gforth  two_tuck
  :
 swap 2over ;
  \ toggle is high-level: 0.11/0.42%
- @       a_addr -- w             core    fetch
+ \g memory
+ @       ( a_addr -- w )         core    fetch
+ ""@i{w} is the cell stored at @i{a_addr}.""
+ w = *a_addr;
+ \ lit@ / lit_fetch = lit @
+ lit@            ( #a_addr -- w ) new    lit_fetch
  w = *a_addr;
- !       w a_addr --             core    store
+ !       ( w a_addr -- )         core    store
+ ""Store @i{w} into the cell at @i{a-addr}.""
  *a_addr = w;
- +!      n a_addr --             core    plus_store
+ +!      ( n a_addr -- )         core    plus_store
+ ""Add @i{n} to the cell at @i{a-addr}.""
  *a_addr += n;
  :
   tuck @ + swap ! ;
- c@      c_addr -- c             core    cfetch
+ c@      ( c_addr -- c )         core    c_fetch
+ ""@i{c} is the char stored at @i{c_addr}.""
  c = *c_addr;
  :
  [ bigendian [IF] ]
- Line 1104
+ Line 1452
  Line 1104
  Line 1452
  ;
  : 8>> 2/ 2/ 2/ 2/  2/ 2/ 2/ 2/ ;
- c!      c c_addr --             core    cstore
+ c!      ( c c_addr -- )         core    c_store
+ ""Store @i{c} into the char at @i{c-addr}.""
  *c_addr = c;
  :
  [ bigendian [IF] ]
- Line 1133
+ Line 1482
  Line 1133
  Line 1482
  [THEN]
  : 8<< 2* 2* 2* 2*  2* 2* 2* 2* ;
-!      w1 w2 a_addr --         core    two_store
+!      ( w1 w2 a_addr -- )             core    two_store
+ ""Store @i{w2} into the cell at @i{c-addr} and @i{w1} into the next cell.""
  a_addr[0] = w2;
  a_addr[1] = w1;
  :
   tuck ! cell+ ! ;
-@      a_addr -- w1 w2         core    two_fetch
+@      ( a_addr -- w1 w2 )             core    two_fetch
+ ""@i{w2} is the content of the cell stored at @i{a-addr}, @i{w1} is
+ the content of the next cell.""
  w2 = a_addr[0];
  w1 = a_addr[1];
  :
   dup cell+ @ swap @ ;
- cell+   a_addr1 -- a_addr2      core    cell_plus
+ cell+   ( a_addr1 -- a_addr2 )  core    cell_plus
+ ""@code{1 cells +}""
  a_addr2 = a_addr1+1;
  :
   cell + ;
- cells   n1 -- n2                core
+ cells   ( n1 -- n2 )            core
+ "" @i{n2} is the number of address units of @i{n1} cells.""
  n2 = n1 * sizeof(Cell);
  :
   [ cell
- Line 1160
+ Line 1514
  Line 1160
  Line 1514
 / dup [IF] ] 2* [ [THEN]
   drop ] ;
- char+   c_addr1 -- c_addr2      core    care_plus
+ char+   ( c_addr1 -- c_addr2 )  core    char_plus
+ ""@code{1 chars +}.""
  c_addr2 = c_addr1 + 1;
  :
 + ;
- (chars)         n1 -- n2        gforth  paren_cares
+ (chars) ( n1 -- n2 )    gforth  paren_chars
  n2 = n1 * sizeof(Char);
  :
   ;
- count   c_addr1 -- c_addr2 u    core
+ count   ( c_addr1 -- c_addr2 u )        core
+ ""@i{c-addr2} is the first character and @i{u} the length of the
+ counted string at @i{c-addr1}.""
  u = *c_addr1;
  c_addr2 = c_addr1+1;
  :
   dup 1+ swap c@ ;
- (f83find)       c_addr u f83name1 -- f83name2   new     paren_f83find
+ \g compiler
- for (; f83name1 != NULL; f83name1 = f83name1->next)
+ \+f83headerstring
+ (f83find)       ( c_addr u f83name1 -- f83name2 )       new     paren_f83find
+ for (; f83name1 != NULL; f83name1 = (struct F83Name *)(f83name1->next))
    if ((UCell)F83NAME_COUNT(f83name1)==u &&
        memcasecmp(c_addr, f83name1->name, u)== 0 /* or inline? */)
      break;
- Line 1190
+ Line 1551
  Line 1190
  Line 1551
      REPEAT  THEN  nip nip ;
  : (find-samelen) ( u f83name1 -- u f83name2/0 )
      BEGIN  2dup cell+ c@ $1F and <> WHILE  @  dup 0= UNTIL  THEN ;
+ : capscomp ( c_addr1 u c_addr2 -- n )
+  swap bounds
+  ?DO  dup c@ I c@ <>
+      IF  dup c@ toupper I c@ toupper =
+      ELSE  true  THEN  WHILE  1+  LOOP  drop 0
+  ELSE  c@ toupper I c@ toupper - unloop  THEN  sgn ;
+ : sgn ( n -- -1/0/1 )
+  dup 0= IF EXIT THEN  0< 2* 1+ ;
- \+has? hash [IF]
+ \-
- (hashfind)      c_addr u a_addr -- f83name2     new     paren_hashfind
+ (listlfind)     ( c_addr u longname1 -- longname2 )     new     paren_listlfind
- F83Name *f83name1;
+ longname2=listlfind(c_addr, u, longname1);
- f83name2=NULL;
+ :
- while(a_addr != NULL)
+     BEGIN  dup WHILE  (findl-samelen)  dup  WHILE
- {
+         >r 2dup r@ cell+ cell+ capscomp  0=
-    f83name1=(F83Name *)(a_addr[1]);
+         IF  2drop r>  EXIT  THEN
-    a_addr=(Cell *)(a_addr[0]);
+         r> @
-    if ((UCell)F83NAME_COUNT(f83name1)==u &&
+     REPEAT  THEN  nip nip ;
-        memcasecmp(c_addr, f83name1->name, u)== 0 /* or inline? */)
+ : (findl-samelen) ( u longname1 -- u longname2/0 )
-      {
+     BEGIN  2dup cell+ @ lcount-mask and <> WHILE  @  dup 0= UNTIL  THEN ;
-         f83name2=f83name1;
+ : capscomp ( c_addr1 u c_addr2 -- n )
-         break;
+  swap bounds
-      }
+  ?DO  dup c@ I c@ <>
- }
+      IF  dup c@ toupper I c@ toupper =
+      ELSE  true  THEN  WHILE  1+  LOOP  drop 0
+  ELSE  c@ toupper I c@ toupper - unloop  THEN  sgn ;
+ : sgn ( n -- -1/0/1 )
+  dup 0= IF EXIT THEN  0< 2* 1+ ;
+ \+hash
+ (hashlfind)     ( c_addr u a_addr -- longname2 )        new     paren_hashlfind
+ longname2 = hashlfind(c_addr, u, a_addr);
  :
   BEGIN  dup  WHILE
-@ >r >r dup r@ cell+ c@ $1F and =
+@ >r >r dup r@ cell+ @ lcount-mask and =
-         IF  2dup r@ cell+ char+ capscomp 0=
+         IF  2dup r@ cell+ cell+ capscomp 0=
              IF  2drop r> rdrop  EXIT  THEN  THEN
          rdrop r>
   REPEAT nip nip ;
- (tablefind)     c_addr u a_addr -- f83name2     new     paren_tablefind
+ (tablelfind)    ( c_addr u a_addr -- longname2 )        new     paren_tablelfind
  ""A case-sensitive variant of @code{(hashfind)}""
- F83Name *f83name1;
+ longname2 = tablelfind(c_addr, u, a_addr);
- f83name2=NULL;
- while(a_addr != NULL)
- {
-    f83name1=(F83Name *)(a_addr[1]);
-    a_addr=(Cell *)(a_addr[0]);
-    if ((UCell)F83NAME_COUNT(f83name1)==u &&
-        memcmp(c_addr, f83name1->name, u)== 0 /* or inline? */)
-      {
-         f83name2=f83name1;
-         break;
-      }
- }
  :
   BEGIN  dup  WHILE
-@ >r >r dup r@ cell+ c@ $1F and =
+@ >r >r dup r@ cell+ @ lcount-mask and =
-         IF  2dup r@ cell+ char+ -text 0=
+         IF  2dup r@ cell+ cell+ -text 0=
              IF  2drop r> rdrop  EXIT  THEN  THEN
          rdrop r>
   REPEAT nip nip ;
+ : -text ( c_addr1 u c_addr2 -- n )
+  swap bounds
+  ?DO  dup c@ I c@ = WHILE  1+  LOOP  drop 0
+  ELSE  c@ I c@ - unloop  THEN  sgn ;
+ : sgn ( n -- -1/0/1 )
+  dup 0= IF EXIT THEN  0< 2* 1+ ;
- (hashkey)       c_addr u1 -- u2         gforth  paren_hashkey
+ (hashkey1)      ( c_addr u ubits -- ukey )              gforth  paren_hashkey1
- u2=0;
- while(u1--)
-    u2+=(Cell)toupper(*c_addr++);
- :
--rot bounds ?DO  I c@ toupper +  LOOP ;
- (hashkey1)      c_addr u ubits -- ukey          gforth  paren_hashkey1
  ""ukey is the hash key for the string c_addr u fitting in ubits bits""
- /* this hash function rotates the key at every step by rot bits within
+ ukey = hashkey1(c_addr, u, ubits);
-    ubits bits and xors it with the character. This function does ok in
-    the chi-sqare-test.  Rot should be <=7 (preferably <=5) for
-    ASCII strings (larger if ubits is large), and should share no
-    divisors with ubits.
- */
- unsigned rot = ((char []){5,0,1,2,3,4,5,5,5,5,3,5,5,5,5,7,5,5,5,5,7,5,5,5,5,6,5,5,5,5,7,5,5})[ubits];
- Char *cp = c_addr;
- for (ukey=0; cp<c_addr+u; cp++)
-     ukey = ((((ukey<<rot) | (ukey>>(ubits-rot)))
-              ^ toupper(*cp))
-             & ((1<<ubits)-1));
  :
   dup rot-values + c@ over 1 swap lshift 1- >r
   tuck - 2swap r> 0 2swap bounds
- Line 1272
+ Line 1626
  Line 1272
  Line 1626
 c, 5 c, 5 c, 5 c, 5 c,  6 c, 5 c, 5 c, 5 c, 5 c,
 c, 5 c, 5 c,
- \+[THEN]
+ \+
- (parse-white)   c_addr1 u1 -- c_addr2 u2        gforth  paren_parse_white
+ \+
- /* use !isgraph instead of isspace? */
- Char *endp = c_addr1+u1;
+ (parse-white)   ( c_addr1 u1 -- c_addr2 u2 )    gforth  paren_parse_white
- while (c_addr1<endp && isspace(*c_addr1))
+ struct Cellpair r=parse_white(c_addr1, u1);
-   c_addr1++;
+ c_addr2 = (Char *)(r.n1);
- if (c_addr1<endp) {
+ u2 = r.n2;
-   for (c_addr2 = c_addr1; c_addr1<endp && !isspace(*c_addr1); c_addr1++)
-     ;
-   u2 = c_addr1-c_addr2;
- }
- else {
-   c_addr2 = c_addr1;
-   u2 = 0;
- }
  :
   BEGIN  dup  WHILE  over c@ bl <=  WHILE  1 /string
   REPEAT  THEN  2dup
   BEGIN  dup  WHILE  over c@ bl >   WHILE  1 /string
   REPEAT  THEN  nip - ;
- aligned         c_addr -- a_addr        core
+ aligned ( c_addr -- a_addr )    core
+ "" @i{a-addr} is the first aligned address greater than or equal to @i{c-addr}.""
  a_addr = (Cell *)((((Cell)c_addr)+(sizeof(Cell)-1))&(-sizeof(Cell)));
  :
   [ cell 1- ] Literal + [ -1 cells ] Literal and ;
- faligned        c_addr -- f_addr        float   f_aligned
+ faligned        ( c_addr -- f_addr )    float   f_aligned
+ "" @i{f-addr} is the first float-aligned address greater than or equal to @i{c-addr}.""
  f_addr = (Float *)((((Cell)c_addr)+(sizeof(Float)-1))&(-sizeof(Float)));
  :
   [ 1 floats 1- ] Literal + [ -1 floats ] Literal and ;
- >body           xt -- a_addr    core    to_body
+ \ threading stuff is currently only interesting if we have a compiler
- a_addr = PFA(xt);
+ \fhas? standardthreading has? compiler and [IF]
- :
+ threading-method        ( -- n )        gforth  threading_method
-cells + ;
- >code-address           xt -- c_addr            gforth  to_code_address
- ""c_addr is the code address of the word xt""
- /* !! This behaves installation-dependently for DOES-words */
- c_addr = (Address)CODE_ADDRESS(xt);
- :
-     @ ;
- >does-code      xt -- a_addr            gforth  to_does_code
- ""If xt ist the execution token of a defining-word-defined word,
- a_addr is the start of the Forth code after the DOES>;
- Otherwise a_addr is 0.""
- a_addr = (Cell *)DOES_CODE(xt);
- :
-     cell+ @ ;
- code-address!           c_addr xt --            gforth  code_address_store
- ""Creates a code field with code address c_addr at xt""
- MAKE_CF(xt, c_addr);
- CACHE_FLUSH(xt,PFA(0));
- :
-     ! ;
- does-code!      a_addr xt --            gforth  does_code_store
- ""creates a code field at xt for a defining-word-defined word; a_addr
- is the start of the Forth code after DOES>""
- MAKE_DOES_CF(xt, a_addr);
- CACHE_FLUSH(xt,PFA(0));
- :
-     dodoes: over ! cell+ ! ;
- does-handler!   a_addr --       gforth  does_handler_store
- ""creates a DOES>-handler at address a_addr. a_addr usually points
- just behind a DOES>.""
- MAKE_DOES_HANDLER(a_addr);
- CACHE_FLUSH(a_addr,DOES_HANDLER_SIZE);
- :
-     drop ;
- /does-handler   -- n    gforth  slash_does_handler
- ""the size of a does-handler (includes possible padding)""
- /* !! a constant or environmental query might be better */
- n = DOES_HANDLER_SIZE;
- :
-cells ;
- threading-method        -- n    gforth  threading_method
  ""0 if the engine is direct threaded. Note that this may change during
  the lifetime of an image.""
  #if defined(DOUBLY_INDIRECT)
- Line 1369
+ Line 1669
  Line 1369
  Line 1669
  :
 ;
- \+has? os [IF]
+ \f[THEN]
+ \g hostos
- (key)   -- n            gforth  paren_key
+ key-file        ( wfileid -- c )                gforth  paren_key_file
+ ""Read one character @i{c} from @i{wfileid}.  This word disables
+ buffering for @i{wfileid}.  If you want to read characters from a
+ terminal in non-canonical (raw) mode, you have to put the terminal in
+ non-canonical mode yourself (using the C interface); the exception is
+ @code{stdin}: Gforth automatically puts it into non-canonical mode.""
+ #ifdef HAS_FILE
  fflush(stdout);
- /* !! noecho */
+ c = key((FILE*)wfileid);
- n = key();
+ #else
+ c = key(stdin);
+ #endif
- key?    -- n            facility        key_q
+ key?-file       ( wfileid -- f )                gforth  key_q_file
+ ""@i{f} is true if at least one character can be read from @i{wfileid}
+ without blocking.  If you also want to use @code{read-file} or
+ @code{read-line} on the file, you have to call @code{key?-file} or
+ @code{key-file} first (these two words disable buffering).""
+ #ifdef HAS_FILE
  fflush(stdout);
- n = key_query;
+ f = key_query((FILE*)wfileid);
+ #else
+ f = key_query(stdin);
+ #endif
+ \+os
+ stdin   ( -- wfileid )  gforth
+ ""The standard input file of the Gforth process.""
+ wfileid = (Cell)stdin;
- stdout  -- wfileid      gforth
+ stdout  ( -- wfileid )  gforth
+ ""The standard output file of the Gforth process.""
  wfileid = (Cell)stdout;
- stderr  -- wfileid      gforth
+ stderr  ( -- wfileid )  gforth
+ ""The standard error output file of the Gforth process.""
  wfileid = (Cell)stderr;
- form    -- urows ucols  gforth
+ form    ( -- urows ucols )      gforth
  ""The number of lines and columns in the terminal. These numbers may change
  with the window size.""
  /* we could block SIGWINCH here to get a consistent size, but I don't
- Line 1394
+ Line 1720
  Line 1394
  Line 1720
  urows=rows;
  ucols=cols;
- flush-icache    c_addr u --     gforth  flush_icache
+ wcwidth ( u -- n )      gforth
+ ""The number of fixed-width characters per unicode character u""
+ n = wcwidth(u);
+ flush-icache    ( c_addr u -- ) gforth  flush_icache
  ""Make sure that the instruction cache of the processor (if there is
- one) does not contain stale data at @var{c_addr} and @var{u} bytes
+ one) does not contain stale data at @i{c-addr} and @i{u} bytes
  afterwards. @code{END-CODE} performs a @code{flush-icache}
  automatically. Caveat: @code{flush-icache} might not work on your
  installation; this is usually the case if direct threading is not
- Line 1406
+ Line 1736
  Line 1406
  Line 1736
  cache.""
  FLUSH_ICACHE(c_addr,u);
- (bye)   n --    gforth  paren_bye
+ (bye)   ( n -- )        gforth  paren_bye
+ SUPER_END;
  return (Label *)n;
- (system)        c_addr u -- wretval wior        gforth  peren_system
+ (system)        ( c_addr u -- wretval wior )    gforth  paren_system
- int old_tp=terminal_prepped;
+ wretval = gforth_system(c_addr, u);
- deprep_terminal();
- wretval=system(cstr(c_addr,u,1)); /* ~ expansion on first part of string? */
  wior = IOR(wretval==-1 || (wretval==127 && errno != 0));
- if (old_tp)
-   prep_terminal();
- getenv  c_addr1 u1 -- c_addr2 u2        gforth
+ getenv  ( c_addr1 u1 -- c_addr2 u2 )    gforth
- c_addr2 = getenv(cstr(c_addr1,u1,1));
+ ""The string @i{c-addr1 u1} specifies an environment variable. The string @i{c-addr2 u2}
- u2 = (c_addr2 == NULL ? 0 : strlen(c_addr2));
+ is the host operating system's expansion of that environment variable. If the
+ environment variable does not exist, @i{c-addr2 u2} specifies a string 0 characters
+ in length.""
+ /* close ' to keep fontify happy */
+ c_addr2 = (Char *)getenv(cstr(c_addr1,u1,1));
+ u2 = (c_addr2 == NULL ? 0 : strlen((char *)c_addr2));
- open-pipe       c_addr u ntype -- wfileid wior  gforth  open_pipe
+ open-pipe       ( c_addr u wfam -- wfileid wior )       gforth  open_pipe
- wfileid=(Cell)popen(cstr(c_addr,u,1),fileattr[ntype]); /* ~ expansion of 1st arg? */
+ wfileid=(Cell)popen(cstr(c_addr,u,1),pfileattr[wfam]); /* ~ expansion of 1st arg? */
  wior = IOR(wfileid==0); /* !! the man page says that errno is not set reliably */
- close-pipe      wfileid -- wretval wior         gforth  close_pipe
+ close-pipe      ( wfileid -- wretval wior )             gforth  close_pipe
  wretval = pclose((FILE *)wfileid);
  wior = IOR(wretval==-1);
- time&date       -- nsec nmin nhour nday nmonth nyear    facility-ext    time_and_date
+ time&date       ( -- nsec nmin nhour nday nmonth nyear )        facility-ext    time_and_date
+ ""Report the current time of day. Seconds, minutes and hours are numbered from 0.
+ Months are numbered from 1.""
+ #if 1
+ time_t now;
+ struct tm *ltime;
+ time(&now);
+ ltime=localtime(&now);
+ #else
  struct timeval time1;
  struct timezone zone1;
  struct tm *ltime;
  gettimeofday(&time1,&zone1);
+ /* !! Single Unix specification:
+    If tzp is not a null pointer, the behaviour is unspecified. */
  ltime=localtime((time_t *)&time1.tv_sec);
+ #endif
  nyear =ltime->tm_year+1900;
  nmonth=ltime->tm_mon+1;
  nday  =ltime->tm_mday;
- Line 1442
+ Line 1785
  Line 1442
  Line 1785
  nmin  =ltime->tm_min;
  nsec  =ltime->tm_sec;
- ms      n --    facility-ext
+ ms      ( n -- )        facility-ext
+ ""Wait at least @i{n} milli-second.""
  struct timeval timeout;
  timeout.tv_sec=n/1000;
  timeout.tv_usec=1000*(n%1000);
  (void)select(0,0,0,0,&timeout);
- allocate        u -- a_addr wior        memory
+ allocate        ( u -- a_addr wior )    memory
+ ""Allocate @i{u} address units of contiguous data space. The initial
+ contents of the data space is undefined. If the allocation is successful,
+ @i{a-addr} is the start address of the allocated region and @i{wior}
+ is 0. If the allocation fails, @i{a-addr} is undefined and @i{wior}
+ is a non-zero I/O result code.""
  a_addr = (Cell *)malloc(u?u:1);
  wior = IOR(a_addr==NULL);
- free            a_addr -- wior          memory
+ free    ( a_addr -- wior )              memory
+ ""Return the region of data space starting at @i{a-addr} to the system.
+ The region must originally have been obtained using @code{allocate} or
+ @code{resize}. If the operational is successful, @i{wior} is 0.
+ If the operation fails, @i{wior} is a non-zero I/O result code.""
  free(a_addr);
  wior = 0;
- resize          a_addr1 u -- a_addr2 wior       memory
+ resize  ( a_addr1 u -- a_addr2 wior )   memory
- ""Change the size of the allocated area at @i{a_addr1} to @i{u}
+ ""Change the size of the allocated area at @i{a-addr1} to @i{u}
  address units, possibly moving the contents to a different
- area. @i{a_addr2} is the address of the resulting area. If
+ area. @i{a-addr2} is the address of the resulting area.
- @code{a_addr1} is 0, Gforth's (but not the standard) @code{resize}
+ If the operation is successful, @i{wior} is 0.
- @code{allocate}s @i{u} address units.""
+ If the operation fails, @i{wior} is a non-zero
+ I/O result code. If @i{a-addr1} is 0, Gforth's (but not the Standard)
+ @code{resize} @code{allocate}s @i{u} address units.""
  /* the following check is not necessary on most OSs, but it is needed
     on SunOS 4.1.2. */
+ /* close ' to keep fontify happy */
  if (a_addr1==NULL)
    a_addr2 = (Cell *)malloc(u);
  else
    a_addr2 = (Cell *)realloc(a_addr1, u);
  wior = IOR(a_addr2==NULL);      /* !! Define a return code */
- strerror        n -- c_addr u   gforth
+ strerror        ( n -- c_addr u )       gforth
- c_addr = strerror(n);
+ c_addr = (Char *)strerror(n);
- u = strlen(c_addr);
+ u = strlen((char *)c_addr);
- strsignal       n -- c_addr u   gforth
+ strsignal       ( n -- c_addr u )       gforth
- c_addr = strsignal(n);
+ c_addr = (Char *)strsignal(n);
- u = strlen(c_addr);
+ u = strlen((char *)c_addr);
- call-c  w --    gforth  call_c
+ call-c  ( ... w -- ... )        gforth  call_c
  ""Call the C function pointed to by @i{w}. The C function has to
  access the stack itself. The stack pointers are exported in the global
  variables @code{SP} and @code{FP}.""
  /* This is a first attempt at support for calls to C. This may change in
     the future */
- IF_FTOS(fp[0]=FTOS);
+ gforth_FP=fp;
- FP=fp;
+ gforth_SP=sp;
- SP=sp;
  ((void (*)())w)();
- sp=SP;
+ sp=gforth_SP;
- fp=FP;
+ fp=gforth_FP;
- IF_TOS(TOS=sp[0]);
- IF_FTOS(FTOS=fp[0]);
- \+[THEN] ( has? os ) has? file [IF]
+ \+
+ \+file
- close-file      wfileid -- wior         file    close_file
+ close-file      ( wfileid -- wior )             file    close_file
  wior = IOR(fclose((FILE *)wfileid)==EOF);
- open-file       c_addr u ntype -- w2 wior       file    open_file
+ open-file       ( c_addr u wfam -- wfileid wior )       file    open_file
- w2 = (Cell)fopen(tilde_cstr(c_addr, u, 1), fileattr[ntype]);
+ wfileid = (Cell)fopen(tilde_cstr(c_addr, u, 1), fileattr[wfam]);
- wior =  IOR(w2 == 0);
+ wior =  IOR(wfileid == 0);
- create-file     c_addr u ntype -- w2 wior       file    create_file
+ create-file     ( c_addr u wfam -- wfileid wior )       file    create_file
  Cell    fd;
- fd = open(tilde_cstr(c_addr, u, 1), O_CREAT|O_TRUNC|ufileattr[ntype], 0666);
+ fd = open(tilde_cstr(c_addr, u, 1), O_CREAT|O_TRUNC|ufileattr[wfam], 0666);
  if (fd != -1) {
-   w2 = (Cell)fdopen(fd, fileattr[ntype]);
+   wfileid = (Cell)fdopen(fd, fileattr[wfam]);
-   wior = IOR(w2 == 0);
+   wior = IOR(wfileid == 0);
  } else {
-   w2 = 0;
+   wfileid = 0;
    wior = IOR(1);
  }
- delete-file     c_addr u -- wior                file    delete_file
+ delete-file     ( c_addr u -- wior )            file    delete_file
  wior = IOR(unlink(tilde_cstr(c_addr, u, 1))==-1);
- rename-file     c_addr1 u1 c_addr2 u2 -- wior   file-ext        rename_file
+ rename-file     ( c_addr1 u1 c_addr2 u2 -- wior )       file-ext        rename_file
- char *s1=tilde_cstr(c_addr2, u2, 1);
+ ""Rename file @i{c_addr1 u1} to new name @i{c_addr2 u2}""
- wior = IOR(rename(tilde_cstr(c_addr1, u1, 0), s1)==-1);
+ wior = rename_file(c_addr1, u1, c_addr2, u2);
- file-position   wfileid -- ud wior      file    file_position
+ file-position   ( wfileid -- ud wior )  file    file_position
  /* !! use tell and lseek? */
- ud = LONG2UD(ftell((FILE *)wfileid));
+ ud = OFF2UD(ftello((FILE *)wfileid));
- wior = IOR(UD2LONG(ud)==-1);
+ wior = IOR(UD2OFF(ud)==-1);
- reposition-file ud wfileid -- wior      file    reposition_file
+ reposition-file ( ud wfileid -- wior )  file    reposition_file
- wior = IOR(fseek((FILE *)wfileid, UD2LONG(ud), SEEK_SET)==-1);
+ wior = IOR(fseeko((FILE *)wfileid, UD2OFF(ud), SEEK_SET)==-1);
- file-size       wfileid -- ud wior      file    file_size
+ file-size       ( wfileid -- ud wior )  file    file_size
- #include <sys/stat.h>
  struct stat buf;
  wior = IOR(fstat(fileno((FILE *)wfileid), &buf)==-1);
- ud = LONG2UD(buf.st_size);
+ ud = OFF2UD(buf.st_size);
- resize-file     ud wfileid -- wior      file    resize_file
+ resize-file     ( ud wfileid -- wior )  file    resize_file
- wior = IOR(ftruncate(fileno((FILE *)wfileid), UD2LONG(ud))==-1);
+ wior = IOR(ftruncate(fileno((FILE *)wfileid), UD2OFF(ud))==-1);
- read-file       c_addr u1 wfileid -- u2 wior    file    read_file
+ read-file       ( c_addr u1 wfileid -- u2 wior )        file    read_file
  /* !! fread does not guarantee enough */
  u2 = fread(c_addr, sizeof(Char), u1, (FILE *)wfileid);
  wior = FILEIO(u2<u1 && ferror((FILE *)wfileid));
- Line 1545
+ Line 1898
  Line 1545
  Line 1898
  if (wior)
    clearerr((FILE *)wfileid);
- read-line       c_addr u1 wfileid -- u2 flag wior       file    read_line
+ (read-line)     ( c_addr u1 wfileid -- u2 flag u3 wior ) file   paren_read_line
- /*
+ struct Cellquad r = read_line(c_addr, u1, wfileid);
- Cell c;
+ u2   = r.n1;
- flag=-1;
+ flag = r.n2;
- for(u2=0; u2<u1; u2++)
+ u3   = r.n3;
- {
+ wior = r.n4;
-    *c_addr++ = (Char)(c = getc((FILE *)wfileid));
-    if(c=='\n') break;
-    if(c==EOF)
-      {
-         flag=FLAG(u2!=0);
-         break;
-      }
- }
- wior=FILEIO(ferror((FILE *)wfileid));
- */
- if ((flag=FLAG(!feof((FILE *)wfileid) &&
-                fgets(c_addr,u1+1,(FILE *)wfileid) != NULL))) {
-   wior=FILEIO(ferror((FILE *)wfileid)); /* !! ior? */
-   if (wior)
-     clearerr((FILE *)wfileid);
-   u2 = strlen(c_addr);
-   u2-=((u2>0) && (c_addr[u2-1]==NEWLINE));
- }
- else {
-   wior=0;
-   u2=0;
- }
- \+[THEN]  has? file [IF] -1 [ELSE] has? os [THEN] [IF]
+ \+
- write-file      c_addr u1 wfileid -- wior       file    write_file
+ write-file      ( c_addr u1 wfileid -- wior )   file    write_file
  /* !! fwrite does not guarantee enough */
+ #ifdef HAS_FILE
  {
    UCell u2 = fwrite(c_addr, sizeof(Char), u1, (FILE *)wfileid);
    wior = FILEIO(u2<u1 && ferror((FILE *)wfileid));
    if (wior)
      clearerr((FILE *)wfileid);
  }
+ #else
+ TYPE(c_addr, u1);
+ #endif
- emit-file       c wfileid -- wior       gforth  emit_file
+ emit-file       ( c wfileid -- wior )   gforth  emit_file
+ #ifdef HAS_FILE
  wior = FILEIO(putc(c, (FILE *)wfileid)==EOF);
  if (wior)
    clearerr((FILE *)wfileid);
+ #else
+ PUTC(c);
+ #endif
- \+[THEN]  has? file [IF]
+ \+file
- flush-file      wfileid -- wior         file-ext        flush_file
+ flush-file      ( wfileid -- wior )             file-ext        flush_file
  wior = IOR(fflush((FILE *) wfileid)==EOF);
- file-status     c_addr u -- ntype wior  file-ext        file_status
+ file-status     ( c_addr u -- wfam wior )       file-ext        file_status
- char *filename=tilde_cstr(c_addr, u, 1);
+ struct Cellpair r = file_status(c_addr, u);
- if (access (filename, F_OK) != 0) {
+ wfam = r.n1;
-   ntype=0;
+ wior = r.n2;
-   wior=IOR(1);
- }
+ file-eof?       ( wfileid -- flag )     gforth  file_eof_query
- else if (access (filename, R_OK | W_OK) == 0) {
+ flag = FLAG(feof((FILE *) wfileid));
-   ntype=2; /* r/w */
-   wior=0;
+ open-dir        ( c_addr u -- wdirid wior )     gforth  open_dir
- }
+ ""Open the directory specified by @i{c-addr, u}
- else if (access (filename, R_OK) == 0) {
+ and return @i{dir-id} for futher access to it.""
-   ntype=0; /* r/o */
+ wdirid = (Cell)opendir(tilde_cstr(c_addr, u, 1));
+ wior =  IOR(wdirid == 0);
+ read-dir        ( c_addr u1 wdirid -- u2 flag wior )    gforth  read_dir
+ ""Attempt to read the next entry from the directory specified
+ by @i{dir-id} to the buffer of length @i{u1} at address @i{c-addr}.
+ If the attempt fails because there is no more entries,
+ @i{ior}=0, @i{flag}=0, @i{u2}=0, and the buffer is unmodified.
+ If the attempt to read the next entry fails because of any other reason,
+ return @i{ior}<>0.
+ If the attempt succeeds, store file name to the buffer at @i{c-addr}
+ and return @i{ior}=0, @i{flag}=true and @i{u2} equal to the size of the file name.
+ If the length of the file name is greater than @i{u1},
+ store first @i{u1} characters from file name into the buffer and
+ indicate "name too long" with @i{ior}, @i{flag}=true, and @i{u2}=@i{u1}.""
+ struct dirent * dent;
+ dent = readdir((DIR *)wdirid);
    wior=0;
+ flag = -1;
+ if(dent == NULL) {
+   u2 = 0;
+   flag = 0;
+ } else {
+   u2 = strlen((char *)dent->d_name);
+   if(u2 > u1) {
+     u2 = u1;
+     wior = -512-ENAMETOOLONG;
  }
- else if (access (filename, W_OK) == 0) {
+   memmove(c_addr, dent->d_name, u2);
-   ntype=4; /* w/o */
-   wior=0;
  }
- else {
-   ntype=1; /* well, we cannot access the file, but better deliver a legal
+ close-dir       ( wdirid -- wior )      gforth  close_dir
-             access mode (r/o bin), so we get a decent error later upon open. */
+ ""Close the directory specified by @i{dir-id}.""
-   wior=0;
+ wior = IOR(closedir((DIR *)wdirid));
+ filename-match  ( c_addr1 u1 c_addr2 u2 -- flag )       gforth  match_file
+ char * string = cstr(c_addr1, u1, 1);
+ char * pattern = cstr(c_addr2, u2, 0);
+ flag = FLAG(!fnmatch(pattern, string, 0));
+ set-dir ( c_addr u -- wior )    gforth set_dir
+ ""Change the current directory to @i{c-addr, u}.
+ Return an error if this is not possible""
+ wior = IOR(chdir(tilde_cstr(c_addr, u, 1)));
+ get-dir ( c_addr1 u1 -- c_addr2 u2 )    gforth get_dir
+ ""Store the current directory in the buffer specified by @{c-addr1, u1}.
+ If the buffer size is not sufficient, return 0 0""
+ c_addr2 = (Char *)getcwd((char *)c_addr1, u1);
+ if(c_addr2 != NULL) {
+   u2 = strlen((char *)c_addr2);
+ } else {
+   u2 = 0;
  }
- \+[THEN] ( has? file ) has? floating [IF]
+ \+
+ newline ( -- c_addr u ) gforth
+ ""String containing the newline sequence of the host OS""
+ char newline[] = {
+ #if DIRSEP=='/'
+ /* Unix */
+ '\n'
+ #else
+ /* DOS, Win, OS/2 */
+ '\r','\n'
+ #endif
+ };
+ c_addr=(Char *)newline;
+ u=sizeof(newline);
+ :
+  "newline count ;
+ Create "newline e? crlf [IF] 2 c, $0D c, [ELSE] 1 c, [THEN] $0A c,
+ \+os
+ utime   ( -- dtime )    gforth
+ ""Report the current time in microseconds since some epoch.""
+ struct timeval time1;
+ gettimeofday(&time1,NULL);
+ dtime = timeval2us(&time1);
+ cputime ( -- duser dsystem ) gforth
+ ""duser and dsystem are the respective user- and system-level CPU
+ times used since the start of the Forth system (excluding child
+ processes), in microseconds (the granularity may be much larger,
+ however).  On platforms without the getrusage call, it reports elapsed
+ time (since some epoch) for duser and 0 for dsystem.""
+ #ifdef HAVE_GETRUSAGE
+ struct rusage usage;
+ getrusage(RUSAGE_SELF, &usage);
+ duser = timeval2us(&usage.ru_utime);
+ dsystem = timeval2us(&usage.ru_stime);
+ #else
+ struct timeval time1;
+ gettimeofday(&time1,NULL);
+ duser = timeval2us(&time1);
+ dsystem = DZERO;
+ #endif
+ \+
+ \+floating
+ \g floating
  comparisons(f, r1 r2, f_, r1, r2, gforth, gforth, float, gforth)
  comparisons(f0, r, f_zero_, r, 0., float, gforth, float, gforth)
- d>f             d -- r          float   d_to_f
+ s>f     ( n -- r )              float   s_to_f
- #ifdef BUGGY_LONG_LONG
+ r = n;
+ d>f     ( d -- r )              float   d_to_f
+ #ifdef BUGGY_LL_D2F
  extern double ldexp(double x, int exp);
- r = ldexp((Float)d.hi,CELL_BITS) + (Float)d.lo;
+ if (DHI(d)<0) {
+ #ifdef BUGGY_LL_ADD
+   DCell d2=dnegate(d);
+ #else
+   DCell d2=-d;
+ #endif
+   r = -(ldexp((Float)DHI(d2),CELL_BITS) + (Float)DLO(d2));
+ } else
+   r = ldexp((Float)DHI(d),CELL_BITS) + (Float)DLO(d);
  #else
  r = d;
  #endif
- f>d             r -- d          float   f_to_d
+ f>d     ( r -- d )              float   f_to_d
- #ifdef BUGGY_LONG_LONG
+ extern DCell double2ll(Float r);
- d.hi = ldexp(r,-CELL_BITS) - (r<0);
+ d = double2ll(r);
- d.lo = r-ldexp((Float)d.hi,CELL_BITS);
- #else
+ f>s     ( r -- n )              float   f_to_s
- d = r;
+ n = (Cell)r;
- #endif
- f!              r f_addr --     float   f_store
+ f!      ( r f_addr -- ) float   f_store
+ ""Store @i{r} into the float at address @i{f-addr}.""
  *f_addr = r;
- f@              f_addr -- r     float   f_fetch
+ f@      ( f_addr -- r ) float   f_fetch
+ ""@i{r} is the float at address @i{f-addr}.""
  r = *f_addr;
- df@             df_addr -- r    float-ext       d_f_fetch
+ df@     ( df_addr -- r )        float-ext       d_f_fetch
+ ""Fetch the double-precision IEEE floating-point value @i{r} from the address @i{df-addr}.""
  #ifdef IEEE_FP
  r = *df_addr;
  #else
  !! df@
  #endif
- df!             r df_addr --    float-ext       d_f_store
+ df!     ( r df_addr -- )        float-ext       d_f_store
+ ""Store @i{r} as double-precision IEEE floating-point value to the
+ address @i{df-addr}.""
  #ifdef IEEE_FP
  *df_addr = r;
  #else
  !! df!
  #endif
- sf@             sf_addr -- r    float-ext       s_f_fetch
+ sf@     ( sf_addr -- r )        float-ext       s_f_fetch
+ ""Fetch the single-precision IEEE floating-point value @i{r} from the address @i{sf-addr}.""
  #ifdef IEEE_FP
  r = *sf_addr;
  #else
  !! sf@
  #endif
- sf!             r sf_addr --    float-ext       s_f_store
+ sf!     ( r sf_addr -- )        float-ext       s_f_store
+ ""Store @i{r} as single-precision IEEE floating-point value to the
+ address @i{sf-addr}.""
  #ifdef IEEE_FP
  *sf_addr = r;
  #else
  !! sf!
  #endif
- f+              r1 r2 -- r3     float   f_plus
+ f+      ( r1 r2 -- r3 ) float   f_plus
  r3 = r1+r2;
- f-              r1 r2 -- r3     float   f_minus
+ f-      ( r1 r2 -- r3 ) float   f_minus
  r3 = r1-r2;
- f*              r1 r2 -- r3     float   f_star
+ f*      ( r1 r2 -- r3 ) float   f_star
  r3 = r1*r2;
- f/              r1 r2 -- r3     float   f_slash
+ f/      ( r1 r2 -- r3 ) float   f_slash
  r3 = r1/r2;
- f**             r1 r2 -- r3     float-ext       f_star_star
+ f**     ( r1 r2 -- r3 ) float-ext       f_star_star
- ""@i{r3} is @i{r1} raised to the @i{r2}th power""
+ ""@i{r3} is @i{r1} raised to the @i{r2}th power.""
  r3 = pow(r1,r2);
- fnegate         r1 -- r2        float
+ fm*     ( r1 n -- r2 )  gforth  fm_star
+ r2 = r1*n;
+ fm/     ( r1 n -- r2 )  gforth  fm_slash
+ r2 = r1/n;
+ fm*/    ( r1 n1 n2 -- r2 )      gforth  fm_star_slash
+ r2 = (r1*n1)/n2;
+ f**2    ( r1 -- r2 )    gforth  fm_square
+ r2 = r1*r1;
+ fnegate ( r1 -- r2 )    float   f_negate
  r2 = - r1;
- fdrop           r --            float
+ fdrop   ( r -- )                float   f_drop
- fdup            r -- r r        float
+ fdup    ( r -- r r )    float   f_dupe
- fswap           r1 r2 -- r2 r1  float
+ fswap   ( r1 r2 -- r2 r1 )      float   f_swap
- fover           r1 r2 -- r1 r2 r1       float
+ fover   ( r1 r2 -- r1 r2 r1 )   float   f_over
- frot            r1 r2 r3 -- r2 r3 r1    float
+ frot    ( r1 r2 r3 -- r2 r3 r1 )        float   f_rote
- fnip            r1 r2 -- r2     gforth
+ fnip    ( r1 r2 -- r2 ) gforth  f_nip
- ftuck           r1 r2 -- r2 r1 r2       gforth
+ ftuck   ( r1 r2 -- r2 r1 r2 )   gforth  f_tuck
- float+          f_addr1 -- f_addr2      float   float_plus
+ float+  ( f_addr1 -- f_addr2 )  float   float_plus
+ ""@code{1 floats +}.""
  f_addr2 = f_addr1+1;
- floats          n1 -- n2        float
+ floats  ( n1 -- n2 )    float
+ ""@i{n2} is the number of address units of @i{n1} floats.""
  n2 = n1*sizeof(Float);
- floor           r1 -- r2        float
+ floor   ( r1 -- r2 )    float
- ""round towards the next smaller integral value, i.e., round toward negative infinity""
+ ""Round towards the next smaller integral value, i.e., round toward negative infinity.""
  /* !! unclear wording */
  r2 = floor(r1);
- fround          r1 -- r2        float
+ fround  ( r1 -- r2 )    gforth  f_round
- ""round to the nearest integral value""
+ ""Round to the nearest integral value.""
- /* !! unclear wording */
- #ifdef HAVE_RINT
  r2 = rint(r1);
- #else
- r2 = floor(r1+0.5);
- /* !! This is not quite true to the rounding rules given in the standard */
- #endif
- fmax            r1 r2 -- r3     float
+ fmax    ( r1 r2 -- r3 ) float   f_max
  if (r1<r2)
    r3 = r2;
  else
    r3 = r1;
- fmin            r1 r2 -- r3     float
+ fmin    ( r1 r2 -- r3 ) float   f_min
  if (r1<r2)
    r3 = r1;
  else
    r3 = r2;
- represent               r c_addr u -- n f1 f2   float
+ represent       ( r c_addr u -- n f1 f2 )       float
  char *sig;
+ size_t siglen;
  int flag;
  int decpt;
  sig=ecvt(r, u, &decpt, &flag);
- n=(r==0 ? 1 : decpt);
+ n=(r==0. ? 1 : decpt);
  f1=FLAG(flag!=0);
- f2=FLAG(isdigit(sig[0])!=0);
+ f2=FLAG(isdigit((unsigned)(sig[0]))!=0);
- memmove(c_addr,sig,u);
+ siglen=strlen((char *)sig);
+ if (siglen>u) /* happens in glibc-2.1.3 if 999.. is rounded up */
+   siglen=u;
+ if (!f2) /* workaround Cygwin trailing 0s for Inf and Nan */
+   for (; sig[siglen-1]=='0'; siglen--);
+     ;
+ memcpy(c_addr,sig,siglen);
+ memset(c_addr+siglen,f2?'0':' ',u-siglen);
- >float  c_addr u -- flag        float   to_float
+ >float  ( c_addr u -- f:... flag )      float   to_float
- /* real signature: c_addr u -- r t / f */
+ ""Actual stack effect: ( c_addr u -- r t | f ).  Attempt to convert the
+ character string @i{c-addr u} to internal floating-point
+ representation. If the string represents a valid floating-point number
+ @i{r} is placed on the floating-point stack and @i{flag} is
+ true. Otherwise, @i{flag} is false. A string of blanks is a special
+ case and represents the floating-point number 0.""
  Float r;
- char *number=cstr(c_addr, u, 1);
+ flag = to_float(c_addr, u, &r);
- char *endconv;
+ if (flag) {
- while(isspace(number[--u]) && u>0);
+   fp--;
- switch(number[u])
+   fp[0]=r;
- {
-    case 'd':
-    case 'D':
-    case 'e':
-    case 'E':  break;
-    default :  u++; break;
- }
- number[u]='\0';
- r=strtod(number,&endconv);
- if((flag=FLAG(!(Cell)*endconv)))
- {
-    IF_FTOS(fp[0] = FTOS);
-    fp += -1;
-    FTOS = r;
- }
- else if(*endconv=='d' || *endconv=='D')
- {
-    *endconv='E';
-    r=strtod(number,&endconv);
-    if((flag=FLAG(!(Cell)*endconv)))
-      {
-         IF_FTOS(fp[0] = FTOS);
-         fp += -1;
-         FTOS = r;
-      }
  }
- fabs            r1 -- r2        float-ext
+ fabs    ( r1 -- r2 )    float-ext       f_abs
  r2 = fabs(r1);
- facos           r1 -- r2        float-ext
+ facos   ( r1 -- r2 )    float-ext       f_a_cos
  r2 = acos(r1);
- fasin           r1 -- r2        float-ext
+ fasin   ( r1 -- r2 )    float-ext       f_a_sine
  r2 = asin(r1);
- fatan           r1 -- r2        float-ext
+ fatan   ( r1 -- r2 )    float-ext       f_a_tan
  r2 = atan(r1);
- fatan2          r1 r2 -- r3     float-ext
+ fatan2  ( r1 r2 -- r3 ) float-ext       f_a_tan_two
- ""@i{r1/r2}=tan@i{r3}. The standard does not require, but probably
+ ""@i{r1/r2}=tan(@i{r3}). ANS Forth does not require, but probably
  intends this to be the inverse of @code{fsincos}. In gforth it is.""
  r3 = atan2(r1,r2);
- fcos            r1 -- r2        float-ext
+ fcos    ( r1 -- r2 )    float-ext       f_cos
  r2 = cos(r1);
- fexp            r1 -- r2        float-ext
+ fexp    ( r1 -- r2 )    float-ext       f_e_x_p
  r2 = exp(r1);
- fexpm1          r1 -- r2        float-ext
+ fexpm1  ( r1 -- r2 )    float-ext       f_e_x_p_m_one
  ""@i{r2}=@i{e}**@i{r1}@minus{}1""
  #ifdef HAVE_EXPM1
  extern double
- Line 1820
+ Line 2264
  Line 1820
  Line 2264
  r2 = exp(r1)-1.;
  #endif
- fln             r1 -- r2        float-ext
+ fln     ( r1 -- r2 )    float-ext       f_l_n
  r2 = log(r1);
- flnp1           r1 -- r2        float-ext
+ flnp1   ( r1 -- r2 )    float-ext       f_l_n_p_one
  ""@i{r2}=ln(@i{r1}+1)""
  #ifdef HAVE_LOG1P
  extern double
- Line 1836
+ Line 2280
  Line 1836
  Line 2280
  r2 = log(r1+1.);
  #endif
- flog            r1 -- r2        float-ext
+ flog    ( r1 -- r2 )    float-ext       f_log
- ""the decimal logarithm""
+ ""The decimal logarithm.""
  r2 = log10(r1);
- falog           r1 -- r2        float-ext
+ falog   ( r1 -- r2 )    float-ext       f_a_log
  ""@i{r2}=10**@i{r1}""
  extern double pow10(double);
  r2 = pow10(r1);
- fsin            r1 -- r2        float-ext
+ fsin    ( r1 -- r2 )    float-ext       f_sine
  r2 = sin(r1);
- fsincos         r1 -- r2 r3     float-ext
+ fsincos ( r1 -- r2 r3 ) float-ext       f_sine_cos
  ""@i{r2}=sin(@i{r1}), @i{r3}=cos(@i{r1})""
  r2 = sin(r1);
  r3 = cos(r1);
- fsqrt           r1 -- r2        float-ext
+ fsqrt   ( r1 -- r2 )    float-ext       f_square_root
  r2 = sqrt(r1);
- ftan            r1 -- r2        float-ext
+ ftan    ( r1 -- r2 )    float-ext       f_tan
  r2 = tan(r1);
  :
   fsincos f/ ;
- fsinh           r1 -- r2        float-ext
+ fsinh   ( r1 -- r2 )    float-ext       f_cinch
  r2 = sinh(r1);
  :
   fexpm1 fdup fdup 1. d>f f+ f/ f+ f2/ ;
- fcosh           r1 -- r2        float-ext
+ fcosh   ( r1 -- r2 )    float-ext       f_cosh
  r2 = cosh(r1);
  :
   fexp fdup 1/f f+ f2/ ;
- ftanh           r1 -- r2        float-ext
+ ftanh   ( r1 -- r2 )    float-ext       f_tan_h
  r2 = tanh(r1);
  :
   f2* fexpm1 fdup 2. d>f f+ f/ ;
- fasinh          r1 -- r2        float-ext
+ fasinh  ( r1 -- r2 )    float-ext       f_a_cinch
  r2 = asinh(r1);
  :
   fdup fdup f* 1. d>f f+ fsqrt f/ fatanh ;
- facosh          r1 -- r2        float-ext
+ facosh  ( r1 -- r2 )    float-ext       f_a_cosh
  r2 = acosh(r1);
  :
   fdup fdup f* 1. d>f f- fsqrt f+ fln ;
- fatanh          r1 -- r2        float-ext
+ fatanh  ( r1 -- r2 )    float-ext       f_a_tan_h
  r2 = atanh(r1);
  :
   fdup f0< >r fabs 1. d>f fover f- f/  f2* flnp1 f2/
   r> IF  fnegate  THEN ;
- sfloats         n1 -- n2        float-ext       s_floats
+ sfloats ( n1 -- n2 )    float-ext       s_floats
+ ""@i{n2} is the number of address units of @i{n1}
+ single-precision IEEE floating-point numbers.""
  n2 = n1*sizeof(SFloat);
- dfloats         n1 -- n2        float-ext       d_floats
+ dfloats ( n1 -- n2 )    float-ext       d_floats
+ ""@i{n2} is the number of address units of @i{n1}
+ double-precision IEEE floating-point numbers.""
  n2 = n1*sizeof(DFloat);
- sfaligned       c_addr -- sf_addr       float-ext       s_f_aligned
+ sfaligned       ( c_addr -- sf_addr )   float-ext       s_f_aligned
+ ""@i{sf-addr} is the first single-float-aligned address greater
+ than or equal to @i{c-addr}.""
  sf_addr = (SFloat *)((((Cell)c_addr)+(sizeof(SFloat)-1))&(-sizeof(SFloat)));
  :
   [ 1 sfloats 1- ] Literal + [ -1 sfloats ] Literal and ;
- dfaligned       c_addr -- df_addr       float-ext       d_f_aligned
+ dfaligned       ( c_addr -- df_addr )   float-ext       d_f_aligned
+ ""@i{df-addr} is the first double-float-aligned address greater
+ than or equal to @i{c-addr}.""
  df_addr = (DFloat *)((((Cell)c_addr)+(sizeof(DFloat)-1))&(-sizeof(DFloat)));
  :
   [ 1 dfloats 1- ] Literal + [ -1 dfloats ] Literal and ;
+ v*      ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r ) gforth v_star
+ ""dot-product: r=v1*v2.  The first element of v1 is at f_addr1, the
+ next at f_addr1+nstride1 and so on (similar for v2). Both vectors have
+ ucount elements.""
+ r = v_star(f_addr1, nstride1, f_addr2, nstride2, ucount);
+ :
+  >r swap 2swap swap 0e r> 0 ?DO
+      dup f@ over + 2swap dup f@ f* f+ over + 2swap
+  LOOP 2drop 2drop ;
+ faxpy   ( ra f_x nstridex f_y nstridey ucount -- )      gforth
+ ""vy=ra*vx+vy""
+ faxpy(ra, f_x, nstridex, f_y, nstridey, ucount);
+ :
+  >r swap 2swap swap r> 0 ?DO
+      fdup dup f@ f* over + 2swap dup f@ f+ dup f! over + 2swap
+  LOOP 2drop 2drop fdrop ;
+ \+
  \ The following words access machine/OS/installation-dependent
  \   Gforth internals
  \ !! how about environmental queries DIRECT-THREADED,
  \   INDIRECT-THREADED, TOS-CACHED, FTOS-CACHED, CODEFIELD-DOES */
  \ local variable implementation primitives
- \+[THEN] ( has? floats ) has? glocals [IF]
- @local#         -- w    gforth  fetch_local_number
+ \+glocals
- w = *(Cell *)(lp+(Cell)NEXT_INST);
- INC_IP(1);
+ \g locals
- @local0 -- w    new     fetch_local_zero
+ @local# ( #noffset -- w )       gforth  fetch_local_number
- w = *(Cell *)(lp+0*sizeof(Cell));
+ w = *(Cell *)(lp+noffset);
- @local1 -- w    new     fetch_local_four
+ @local0 ( -- w )        new     fetch_local_zero
- w = *(Cell *)(lp+1*sizeof(Cell));
+ w = ((Cell *)lp)[0];
- @local2 -- w    new     fetch_local_eight
+ @local1 ( -- w )        new     fetch_local_four
- w = *(Cell *)(lp+2*sizeof(Cell));
+ w = ((Cell *)lp)[1];
- @local3 -- w    new     fetch_local_twelve
+ @local2 ( -- w )        new     fetch_local_eight
- w = *(Cell *)(lp+3*sizeof(Cell));
+ w = ((Cell *)lp)[2];
- \+has? floating [IF]
+ @local3 ( -- w )        new     fetch_local_twelve
+ w = ((Cell *)lp)[3];
- f@local#        -- r    gforth  f_fetch_local_number
+ \+floating
- r = *(Float *)(lp+(Cell)NEXT_INST);
- INC_IP(1);
- f@local0        -- r    new     f_fetch_local_zero
+ f@local#        ( #noffset -- r )       gforth  f_fetch_local_number
- r = *(Float *)(lp+0*sizeof(Float));
+ r = *(Float *)(lp+noffset);
- f@local1        -- r    new     f_fetch_local_eight
+ f@local0        ( -- r )        new     f_fetch_local_zero
- r = *(Float *)(lp+1*sizeof(Float));
+ r = ((Float *)lp)[0];
- \+[THEN]
+ f@local1        ( -- r )        new     f_fetch_local_eight
+ r = ((Float *)lp)[1];
- laddr#          -- c_addr       gforth  laddr_number
+ \+
+ laddr#  ( #noffset -- c_addr )  gforth  laddr_number
  /* this can also be used to implement lp@ */
- c_addr = (Char *)(lp+(Cell)NEXT_INST);
+ c_addr = (Char *)(lp+noffset);
- INC_IP(1);
- lp+!#   --      gforth  lp_plus_store_number
+ lp+!#   ( #noffset -- ) gforth  lp_plus_store_number
  ""used with negative immediate values it allocates memory on the
  local stack, a positive immediate argument drops memory from the local
  stack""
- lp += (Cell)NEXT_INST;
+ lp += noffset;
- INC_IP(1);
- lp-     --      new     minus_four_lp_plus_store
+ lp-     ( -- )  new     minus_four_lp_plus_store
  lp += -sizeof(Cell);
- lp+     --      new     eight_lp_plus_store
+ lp+     ( -- )  new     eight_lp_plus_store
  lp += sizeof(Float);
- lp+2    --      new     sixteen_lp_plus_store
+ lp+2    ( -- )  new     sixteen_lp_plus_store
  lp += 2*sizeof(Float);
- lp!     c_addr --       gforth  lp_store
+ lp!     ( c_addr -- )   gforth  lp_store
  lp = (Address)c_addr;
- >l      w --    gforth  to_l
+ >l      ( w -- )        gforth  to_l
  lp -= sizeof(Cell);
  *(Cell *)lp = w;
- \+has? floating [IF]
+ \+floating
- f>l     r --    gforth  f_to_l
+ f>l     ( r -- )        gforth  f_to_l
  lp -= sizeof(Float);
  *(Float *)lp = r;
- \+[THEN]  [THEN] \ has? glocals
+ fpick   ( f:... u -- f:... r )          gforth
+ ""Actually the stack effect is @code{ r0 ... ru u -- r0 ... ru r0 }.""
+ r = fp[u];
+ :
+  floats fp@ + f@ ;
+ \+
+ \+
+ \+OS
+ \g syslib
+ open-lib        ( c_addr1 u1 -- u2 )    gforth  open_lib
+ #if defined(HAVE_LIBDL) || defined(HAVE_DLOPEN)
+ #ifndef RTLD_GLOBAL
+ #define RTLD_GLOBAL 0
+ #endif
+ u2=(UCell) dlopen(cstr(c_addr1, u1, 1), RTLD_GLOBAL | RTLD_LAZY);
+ #else
+ #  ifdef _WIN32
+ u2 = (Cell) GetModuleHandle(cstr(c_addr1, u1, 1));
+ #  else
+ #warning Define open-lib!
+ u2 = 0;
+ #  endif
+ #endif
+ lib-sym ( c_addr1 u1 u2 -- u3 ) gforth  lib_sym
+ #if defined(HAVE_LIBDL) || defined(HAVE_DLOPEN)
+ u3 = (UCell) dlsym((void*)u2,cstr(c_addr1, u1, 1));
+ #else
+ #  ifdef _WIN32
+ u3 = (Cell) GetProcAddress((HMODULE)u2, cstr(c_addr1, u1, 1));
+ #  else
+ #warning Define lib-sym!
+ u3 = 0;
+ #  endif
+ #endif
+ wcall   ( ... u -- ... )        gforth
+ gforth_FP=fp;
+ sp=(Cell*)(SYSCALL(Cell*(*)(Cell *, void *))u)(sp, &gforth_FP);
+ fp=gforth_FP;
+ uw@ ( c_addr -- u )     gforth u_w_fetch
+ ""@i{u} is the zero-extended 16-bit value stored at @i{c_addr}.""
+ u = *(UWyde*)(c_addr);
+ sw@ ( c_addr -- n )     gforth s_w_fetch
+ ""@i{n} is the sign-extended 16-bit value stored at @i{c_addr}.""
+ n = *(Wyde*)(c_addr);
+ w! ( w c_addr -- )      gforth w_store
+ ""Store the bottom 16 bits of @i{w} at @i{c_addr}.""
+ *(Wyde*)(c_addr) = w;
+ ul@ ( c_addr -- u )     gforth u_l_fetch
+ ""@i{u} is the zero-extended 32-bit value stored at @i{c_addr}.""
+ u = *(UTetrabyte*)(c_addr);
+ sl@ ( c_addr -- n )     gforth s_l_fetch
+ ""@i{n} is the sign-extended 32-bit value stored at @i{c_addr}.""
+ n = *(Tetrabyte*)(c_addr);
+ l! ( w c_addr -- )      gforth l_store
+ ""Store the bottom 32 bits of @i{w} at @i{c_addr}.""
+ *(Tetrabyte*)(c_addr) = w;
+ \+FFCALL
+ av-start-void   ( c_addr -- )   gforth  av_start_void
+ av_start_void(alist, c_addr);
+ av-start-int    ( c_addr -- )   gforth  av_start_int
+ av_start_int(alist, c_addr, &irv);
+ av-start-float  ( c_addr -- )   gforth  av_start_float
+ av_start_float(alist, c_addr, &frv);
+ av-start-double ( c_addr -- )   gforth  av_start_double
+ av_start_double(alist, c_addr, &drv);
+ av-start-longlong       ( c_addr -- )   gforth  av_start_longlong
+ av_start_longlong(alist, c_addr, &llrv);
+ av-start-ptr    ( c_addr -- )   gforth  av_start_ptr
+ av_start_ptr(alist, c_addr, void*, &prv);
+ av-int  ( w -- )  gforth  av_int
+ av_int(alist, w);
- \+has? OS [IF]
+ av-float        ( r -- )        gforth  av_float
+ av_float(alist, r);
+ av-double       ( r -- )        gforth  av_double
+ av_double(alist, r);
+ av-longlong     ( d -- )        gforth  av_longlong
+ #ifdef BUGGY_LL_SIZE
+ av_longlong(alist, DLO(d));
+ #else
+ av_longlong(alist, d);
+ #endif
+ av-ptr  ( c_addr -- )   gforth  av_ptr
+ av_ptr(alist, void*, c_addr);
+ av-int-r  ( R:w -- )  gforth  av_int_r
+ av_int(alist, w);
+ av-float-r      ( -- )  gforth  av_float_r
+ float r = *(Float*)lp;
+ lp += sizeof(Float);
+ av_float(alist, r);
+ av-double-r     ( -- )  gforth  av_double_r
+ double r = *(Float*)lp;
+ lp += sizeof(Float);
+ av_double(alist, r);
+ av-longlong-r   ( R:d -- )      gforth  av_longlong_r
+ #ifdef BUGGY_LL_SIZE
+ av_longlong(alist, DLO(d));
+ #else
+ av_longlong(alist, d);
+ #endif
+ av-ptr-r        ( R:c_addr -- ) gforth  av_ptr_r
+ av_ptr(alist, void*, c_addr);
+ av-call-void    ( ... -- ... )  gforth  av_call_void
+ SAVE_REGS
+ av_call(alist);
+ REST_REGS
+ av-call-int     ( ... -- ... w )        gforth  av_call_int
+ SAVE_REGS
+ av_call(alist);
+ REST_REGS
+ w = irv;
+ av-call-float   ( ... -- ... r )        gforth  av_call_float
+ SAVE_REGS
+ av_call(alist);
+ REST_REGS
+ r = frv;
+ av-call-double  ( ... -- ... r )        gforth  av_call_double
+ SAVE_REGS
+ av_call(alist);
+ REST_REGS
+ r = drv;
+ av-call-longlong        ( ... -- ... d )        gforth  av_call_longlong
+ SAVE_REGS
+ av_call(alist);
+ REST_REGS
+ #ifdef BUGGY_LONG_LONG
+ DLO_IS(d, llrv);
+ DHI_IS(d, 0);
+ #else
+ d = llrv;
+ #endif
+ av-call-ptr     ( ... -- ... c_addr )   gforth  av_call_ptr
+ SAVE_REGS
+ av_call(alist);
+ REST_REGS
+ c_addr = prv;
+ alloc-callback  ( a_ip -- c_addr )      gforth  alloc_callback
+ c_addr = (char *)alloc_callback(gforth_callback, (Xt *)a_ip);
+ va-start-void   ( -- )  gforth  va_start_void
+ va_start_void(gforth_clist);
+ va-start-int    ( -- )  gforth  va_start_int
+ va_start_int(gforth_clist);
+ va-start-longlong       ( -- )  gforth  va_start_longlong
+ va_start_longlong(gforth_clist);
+ va-start-ptr    ( -- )  gforth  va_start_ptr
+ va_start_ptr(gforth_clist, (char *));
+ va-start-float  ( -- )  gforth  va_start_float
+ va_start_float(gforth_clist);
+ va-start-double ( -- )  gforth  va_start_double
+ va_start_double(gforth_clist);
+ va-arg-int      ( -- w )        gforth  va_arg_int
+ w = va_arg_int(gforth_clist);
+ va-arg-longlong ( -- d )        gforth  va_arg_longlong
+ #ifdef BUGGY_LONG_LONG
+ DLO_IS(d, va_arg_longlong(gforth_clist));
+ DHI_IS(d, 0);
+ #else
+ d = va_arg_longlong(gforth_clist);
+ #endif
+ va-arg-ptr      ( -- c_addr )   gforth  va_arg_ptr
+ c_addr = (char *)va_arg_ptr(gforth_clist,char*);
+ va-arg-float    ( -- r )        gforth  va_arg_float
+ r = va_arg_float(gforth_clist);
+ va-arg-double   ( -- r )        gforth  va_arg_double
+ r = va_arg_double(gforth_clist);
+ va-return-void ( -- )   gforth va_return_void
+ va_return_void(gforth_clist);
+ return 0;
+ va-return-int ( w -- )  gforth va_return_int
+ va_return_int(gforth_clist, w);
+ return 0;
+ va-return-ptr ( c_addr -- )     gforth va_return_ptr
+ va_return_ptr(gforth_clist, void *, c_addr);
+ return 0;
+ va-return-longlong ( d -- )     gforth va_return_longlong
+ #ifdef BUGGY_LONG_LONG
+ va_return_longlong(gforth_clist, d.lo);
+ #else
+ va_return_longlong(gforth_clist, d);
+ #endif
+ return 0;
+ va-return-float ( r -- )        gforth va_return_float
+ va_return_float(gforth_clist, r);
+ return 0;
+ va-return-double ( r -- )       gforth va_return_double
+ va_return_double(gforth_clist, r);
+ return 0;
+ \+
+ \+LIBFFI
+ ffi-type ( n -- a_type )        gforth ffi_type
+ static void* ffi_types[] =
+     { &ffi_type_void,
+       &ffi_type_uint8, &ffi_type_sint8,
+       &ffi_type_uint16, &ffi_type_sint16,
+       &ffi_type_uint32, &ffi_type_sint32,
+       &ffi_type_uint64, &ffi_type_sint64,
+       &ffi_type_float, &ffi_type_double, &ffi_type_longdouble,
+       &ffi_type_pointer };
+ a_type = ffi_types[n];
+ ffi-size ( n1 -- n2 )   gforth ffi_size
+ static int ffi_sizes[] =
+     { sizeof(ffi_cif), sizeof(ffi_closure) };
+ n2 = ffi_sizes[n1];
+ ffi-prep-cif ( a_atypes n a_rtype a_cif -- w )  gforth ffi_prep_cif
+ w = ffi_prep_cif((ffi_cif *)a_cif, FFI_DEFAULT_ABI, n,
+          (ffi_type *)a_rtype, (ffi_type **)a_atypes);
+ ffi-call ( a_avalues a_rvalue a_ip a_cif -- )   gforth ffi_call
+ SAVE_REGS
+ ffi_call((ffi_cif *)a_cif, (void(*)())a_ip, (void *)a_rvalue, (void **)a_avalues);
+ REST_REGS
+ ffi-prep-closure ( a_ip a_cif a_closure -- w )  gforth ffi_prep_closure
+ w = ffi_prep_closure((ffi_closure *)a_closure, (ffi_cif *)a_cif, gforth_callback, (void *)a_ip);
+ ffi-2@ ( a_addr -- d )  gforth ffi_2fetch
+ #ifdef BUGGY_LONG_LONG
+ DLO_IS(d, *(Cell*)(*a_addr));
+ DHI_IS(d, 0);
+ #else
+ d = *(DCell*)(a_addr);
+ #endif
+ ffi-2! ( d a_addr -- )  gforth ffi_2store
+ #ifdef BUGGY_LONG_LONG
+ *(Cell*)(a_addr) = DLO(d);
+ #else
+ *(DCell*)(a_addr) = d;
+ #endif
+ ffi-arg-int ( -- w )    gforth ffi_arg_int
+ w = *(int *)(*gforth_clist++);
+ ffi-arg-long ( -- w )   gforth ffi_arg_long
+ w = *(long *)(*gforth_clist++);
+ ffi-arg-longlong ( -- d )       gforth ffi_arg_longlong
+ #ifdef BUGGY_LONG_LONG
+ DLO_IS(d, *(Cell*)(*gforth_clist++));
+ DHI_IS(d, -(*(Cell*)(*gforth_clist++)<0));
+ #else
+ d = *(DCell*)(*gforth_clist++);
+ #endif
+ ffi-arg-dlong ( -- d )  gforth ffi_arg_dlong
+ #ifdef BUGGY_LONG_LONG
+ DLO_IS(d, *(Cell*)(*gforth_clist++));
+ DHI_IS(d, -(*(Cell*)(*gforth_clist++)<0));
+ #else
+ d = *(Cell*)(*gforth_clist++);
+ #endif
+ ffi-arg-ptr ( -- c_addr )       gforth ffi_arg_ptr
+ c_addr = *(Char **)(*gforth_clist++);
+ ffi-arg-float ( -- r )  gforth ffi_arg_float
+ r = *(float*)(*gforth_clist++);
+ ffi-arg-double ( -- r ) gforth ffi_arg_double
+ r = *(double*)(*gforth_clist++);
+ ffi-ret-void ( -- )     gforth ffi_ret_void
+ return 0;
+ ffi-ret-int ( w -- )    gforth ffi_ret_int
+ *(int*)(gforth_ritem) = w;
+ return 0;
+ ffi-ret-longlong ( d -- )       gforth ffi_ret_longlong
+ #ifdef BUGGY_LONG_LONG
+ *(Cell*)(gforth_ritem) = DLO(d);
+ #else
+ *(DCell*)(gforth_ritem) = d;
+ #endif
+ return 0;
+ ffi-ret-dlong ( d -- )  gforth ffi_ret_dlong
+ #ifdef BUGGY_LONG_LONG
+ *(Cell*)(gforth_ritem) = DLO(d);
+ #else
+ *(Cell*)(gforth_ritem) = d;
+ #endif
+ return 0;
+ ffi-ret-long ( n -- )   gforth ffi_ret_long
+ *(Cell*)(gforth_ritem) = n;
+ return 0;
+ ffi-ret-ptr ( c_addr -- )       gforth ffi_ret_ptr
+ *(Char **)(gforth_ritem) = c_addr;
+ return 0;
+ ffi-ret-float ( r -- )  gforth ffi_ret_float
+ *(float*)(gforth_ritem) = r;
+ return 0;
+ ffi-ret-double ( r -- ) gforth ffi_ret_double
+ *(double*)(gforth_ritem) = r;
+ return 0;
+ \+
+ \+OLDCALL
  define(`uploop',
         `pushdef(`$1', `$2')_uploop(`$1', `$2', `$3', `$4', `$5')`'popdef(`$1')')
- Line 2003
+ Line 2831
  Line 2003
  Line 2831
                 `uploop(`_i', 1, $1, `format(`u%d, ', _i)', `format(`u%d', _i)')')')
  \ icall(argnum)
  define(icall,
- `icall$1        argflist($1)u -- uret   gforth
+ `icall$1        ( argflist($1)u -- uret )       gforth
- uret = ((Cell(*)(argdlist($1)))u)(argclist($1));
+ uret = (SYSCALL(Cell(*)(argdlist($1)))u)(argclist($1));
  ')
  define(fcall,
- `fcall$1        argflist($1)u -- rret   gforth
+ `fcall$1        ( argflist($1)u -- rret )       gforth
- rret = ((Float(*)(argdlist($1)))u)(argclist($1));
+ rret = (SYSCALL(Float(*)(argdlist($1)))u)(argclist($1));
  ')
+ \ close ' to keep fontify happy
- open-lib        c_addr1 u1 -- u2        gforth  open_lib
+ uploop(i, 0, 7, `icall(i)')
- #if defined(HAVE_LIBDL) || defined(HAVE_DLOPEN)
+ icall(20)
- u2=(UCell) dlopen(cstr(c_addr1, u1, 1), RTLD_LAZY);
+ uploop(i, 0, 7, `fcall(i)')
- #else
+ fcall(20)
- #  ifdef HAVE_LIBKERNEL32
- u2 = (Cell) GetModuleHandle(cstr(c_addr1, u1, 1));
+ \+
- #  else
+ \+
- #warning Define open-lib!
- u2 = 0;
+ \g peephole
- #  endif
+ \+peephole
+ compile-prim1 ( a_prim -- ) gforth compile_prim1
+ ""compile prim (incl. immargs) at @var{a_prim}""
+ compile_prim1(a_prim);
+ finish-code ( ... -- ... ) gforth finish_code
+ ""Perform delayed steps in code generation (branch resolution, I-cache
+ flushing).""
+ /* The ... above are a workaround for a bug in gcc-2.95, which fails
+    to save spTOS (gforth-fast --enable-force-reg) */
+ finish_code();
+ forget-dyncode ( c_code -- f ) gforth-internal forget_dyncode
+ f = forget_dyncode(c_code);
+ decompile-prim ( a_code -- a_prim ) gforth-internal decompile_prim
+ ""a_prim is the code address of the primitive that has been
+ compile_prim1ed to a_code""
+ a_prim = (Cell *)decompile_code((Label)a_code);
+ \ set-next-code and call2 do not appear in images and can be
+ \ renumbered arbitrarily
+ set-next-code ( #w -- ) gforth set_next_code
+ #ifdef NO_IP
+ next_code = (Label)w;
  #endif
- lib-sym c_addr1 u1 u2 -- u3     gforth  lib_sym
+ call2 ( #a_callee #a_ret_addr -- R:a_ret_addr ) gforth
- #if defined(HAVE_LIBDL) || defined(HAVE_DLOPEN)
+ /* call with explicit return address */
- u3 = (UCell) dlsym((void*)u2,cstr(c_addr1, u1, 1));
+ #ifdef NO_IP
- #else
+ INST_TAIL;
- #  ifdef HAVE_LIBKERNEL32
+ JUMP(a_callee);
- u3 = (Cell) GetProcAddress((HMODULE)u2, cstr(c_addr1, u1, 1));
  #  else
- #warning Define lib-sym!
+ assert(0);
- u3 = 0;
- #  endif
  #endif
- uploop(i, 0, 7, `icall(i)')
+ tag-offsets ( -- a_addr ) gforth tag_offsets
- icall(20)
+ extern Cell groups[32];
- uploop(i, 0, 7, `fcall(i)')
+ a_addr = groups;
- fcall(20)
- \+[THEN] \ has? OS
+ \+
- up!     a_addr --       gforth  up_store
+ \g static_super
- UP=up=(char *)a_addr;
- :
+ ifdef(`STACK_CACHE_FILE',
-  up ! ;
+ `include(peeprules.vmg)')
- Variable UP
+ \g end

-Generate output suitable for use with a patch program
+Legend:



Removed from v.1.6
 


changed lines


 
Added in v.1.201
 Legend:



Removed from v.1.6
 


changed lines


 
Added in v.1.201
-Removed from v.1.6
+Added in v.1.201

CVS Admin

gforth: gforth/prim

Diff for /gforth/prim between version 1.6 and 1.201

ViewCVS and CVS Help