Diff for /gforth/prims2x.fs between versions 1.16 and 1.169

version 1.16, 1995/11/07 18:06:57 version 1.169, 2007/12/31 18:40:24
Line 1 Line 1
 \ converts primitives to, e.g., C code   \ converts primitives to, e.g., C code 
   
 \ Copyright (C) 1995 Free Software Foundation, Inc.  \ Copyright (C) 1995,1996,1997,1998,2000,2003,2004,2005,2006,2007 Free Software Foundation, Inc.
   
 \ This file is part of Gforth.  \ This file is part of Gforth.
   
 \ Gforth is free software; you can redistribute it and/or  \ Gforth is free software; you can redistribute it and/or
 \ modify it under the terms of the GNU General Public License  \ modify it under the terms of the GNU General Public License
 \ as published by the Free Software Foundation; either version 2  \ as published by the Free Software Foundation, either version 3
 \ of the License, or (at your option) any later version.  \ of the License, or (at your option) any later version.
   
 \ This program is distributed in the hope that it will be useful,  \ This program is distributed in the hope that it will be useful,
Line 15 Line 15
 \ GNU General Public License for more details.  \ GNU General Public License for more details.
   
 \ You should have received a copy of the GNU General Public License  \ You should have received a copy of the GNU General Public License
 \ along with this program; if not, write to the Free Software  \ along with this program. If not, see http://www.gnu.org/licenses/.
 \ Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.  
   
   
 \ This is not very nice (hard limits, no checking, assumes 1 chars = 1)  \ This is not very nice (hard limits, no checking, assumes 1 chars = 1).
   \ And it grew even worse when it aged.
   
 \ Optimizations:  \ Optimizations:
 \ superfluous stores are removed. GCC removes the superfluous loads by itself  \ superfluous stores are removed. GCC removes the superfluous loads by itself
Line 39 Line 39
 \ add the store optimization for doubles  \ add the store optimization for doubles
 \ regarding problem 1 above: It would be better (for over) to implement  \ regarding problem 1 above: It would be better (for over) to implement
 \       the alternative  \       the alternative
   \ store optimization for combined instructions.
   
 warnings off  \ Design Uglyness:
   
 [IFUNDEF] vocabulary    include search-order.fs [THEN]  
 [IFUNDEF] environment?  include environ.fs      [THEN]  
 include gray.fs  
   
 100 constant max-effect \ number of things on one side of a stack effect  
 4096 constant batch-size \ no meaning, just make sure it's >0  
 255 constant maxchar  
 maxchar 1+ constant eof-char  
 9 constant tab-char  
 10 constant nl-char  
   
 : read-whole-file ( c-addr1 file-id -- c-addr2 )  
 \ reads the contents of the file file-id puts it into memory at c-addr1  
 \ c-addr2 is the first address after the file block  
   begin ( c-addr file-id )  
     2dup batch-size swap read-file   
     if  
       true abort" I/O error"  
     endif  
     ( c-addr file-id actual-size ) rot over + -rot  
     batch-size <>  
   until  
   drop ;  
   
 variable input \ pointer to next character to be parsed  
 variable endinput \ pointer to the end of the input (the char after the last)  
   
 : start ( -- addr )  \ - global state (values, variables) in connection with combined instructions.
  input @ ;  
   
 : end ( addr -- addr u )  
  input @ over - ;  
   
 variable output \ xt ( -- ) of output word  \ - index computation is different for instruction-stream and the
   \ stacks; there are two mechanisms for dealing with that
   \ (stack-in-index-xt and a test for stack==instruction-stream); there
   \ should be only one.
   
 : printprim ( -- )  
  output @ execute ;  
   
 : field  \ for backwards compatibility, jaw
  <builds-field ( n1 n2 -- n3 )  require compat/strcomp.fs
  does>         ( addr1 -- addr2 )  
    @ + ;  
   
 : const-field  
  <builds-field ( n1 n2 -- n3 )  
  does>         ( addr -- w )  
    @ + @ ;  
   
 struct  
  2 cells field item-name  
  cell field item-d-offset  
  cell field item-f-offset  
  cell field item-type  
 constant item-descr  
   
 2variable forth-name  
 2variable wordset  
 2variable c-name  
 2variable doc  
 2variable c-code  
 2variable forth-code  
 2variable stack-string  
 create effect-in  max-effect item-descr * allot  
 create effect-out max-effect item-descr * allot  
 variable effect-in-end ( pointer )  
 variable effect-out-end ( pointer )  
 2variable effect-in-size  
 2variable effect-out-size  
   
 variable primitive-number -10 primitive-number !  
   
 \ for several reasons stack items of a word are stored in a wordlist  
 \ since neither forget nor marker are implemented yet, we make a new  
 \ wordlist for every word and store it in the variable items  
 variable items  
   
 \ a few more set ops  [undefined] outfile-execute [if]
       : outfile-execute ( ... xt file-id -- ... )
           \ unsafe replacement
           outfile-id >r to outfile-id execute r> to outfile-id ;
   [then]
   
 : bit-equivalent ( w1 w2 -- w3 )  warnings off
  xor invert ;  
   
 : complement ( set1 -- set2 )  \ redefinitions of kernel words not present in gforth-0.6.1
  empty ['] bit-equivalent binary-set-operation ;  : latestxt lastcfa @ ;
   : latest last @ ;
   
 \ the parser  [IFUNDEF] try
   include startup.fs
   [THEN]
   
 eof-char max-member \ the whole character set + EOF  : struct% struct ; \ struct is redefined in gray
   
 : getinput ( -- n )  warnings off
  input @  \ warnings on
  dup endinput @ =  
  if  
    drop eof-char  
  else  
    c@  
  endif ;  
   
 :noname ( n -- )  include ./gray.fs
  dup bl > if  128 constant max-effect \ number of things on one side of a stack effect
   emit space  4 constant max-stacks  \ the max. number of stacks (including inst-stream).
  else  255 constant maxchar
   .  maxchar 1+ constant eof-char
  endif ;  #tab constant tab-char
 print-token !  #lf constant nl-char
   
 : testchar? ( set -- f )  variable rawinput \ pointer to next character to be scanned
  getinput member? ;  variable endrawinput \ pointer to the end of the input (the char after the last)
 ' testchar? test-vector !  variable cookedinput \ pointer to the next char to be parsed
   variable line \ line number of char pointed to by input
   variable line-start \ pointer to start of current line (for error messages)
   0 line !
   2variable filename \ filename of original input file
   0 0 filename 2!
   2variable out-filename \ filename of the output file (for sync lines)
   0 0 out-filename 2!
   2variable f-comment
   0 0 f-comment 2!
   variable skipsynclines \ are sync lines ("#line ...") invisible to the parser?
   skipsynclines on
   variable out-nls \ newlines in output (for output sync lines)
   0 out-nls !
   variable store-optimization \ use store optimization?
   store-optimization off
   
   variable include-skipped-insts
   \ does the threaded code for a combined instruction include the cells
   \ for the component instructions (true) or only the cells for the
   \ inline arguments (false)
   include-skipped-insts off
   
   2variable threaded-code-pointer-type \ type used for geninst etc.
   s" Inst **" threaded-code-pointer-type 2!
   
   variable immarg \ values for immediate arguments (to be used in IMM_ARG macros)
   $12340000 immarg !
   
   : th ( addr1 n -- addr2 )
       cells + ;
   
   : holds ( addr u -- )
       \ like HOLD, but for a string
       tuck + swap 0 +do
           1- dup c@ hold
       loop
       drop ;
   
   : insert-wordlist { c-addr u wordlist xt -- }
       \ adds name "addr u" to wordlist using defining word xt
       \ xt may cause additional stack effects
       get-current >r wordlist set-current
       c-addr u nextname xt execute
       r> set-current ;
   
 : ?nextchar ( f -- )  : start ( -- addr )
  ?not? if   cookedinput @ ;
    ." syntax error" cr  
    getinput . cr  
    input @ endinput @ over - 100 min type cr  
    abort  
  endif  
  input @ endinput @ <> if  
    1 input +!  
  endif ;  
   
 : charclass ( set "name" -- )  : end ( addr -- addr u )
  ['] ?nextchar terminal ;   cookedinput @ over - ;
   
 : .. ( c1 c2 -- set )  : print-error-line ( -- )
  ( creates a set that includes the characters c, c1<=c<=c2 )      \ print the current line and position
  empty copy-set      line-start @ endrawinput @ over - 2dup nl-char scan drop nip ( start end )
  swap 1+ rot do      over - type cr
   i over add-member      line-start @ rawinput @ over - typewhite ." ^" cr ;
  loop ;  
   : print-error { addr u -- }
       filename 2@ type ." :" line @ 0 .r ." : " addr u type cr
       print-error-line ;
   
   : ?print-error { f addr u -- }
       f ?not? if
           addr u ['] print-error stderr outfile-execute
           1 (bye) \ abort
       endif ;
   
   : quote ( -- )
       [char] " emit ;
   
   \ count output lines to generate sync lines for output
   
   : count-nls ( addr u -- )
       bounds u+do
           i c@ nl-char = negate out-nls +!
       loop ;
   
   :noname ( addr u -- )
       2dup count-nls
       defers type ;
   is type
   
   variable output          \ xt ( -- ) of output word for simple primitives
   variable output-combined \ xt ( -- ) of output word for combined primitives
   
   struct%
       cell%    field stack-number \ the number of this stack
       cell% 2* field stack-pointer \ stackpointer name
       cell%    field stack-type \ name for default type of stack items
       cell%    field stack-in-index-xt \ ( in-size item -- in-index )
       cell%    field stack-access-transform \ ( nitem -- index )
   end-struct stack%
   
   struct%
    cell% 2* field item-name   \ name, excluding stack prefixes
    cell%    field item-stack  \ descriptor for the stack used, 0 is default
    cell%    field item-type   \ descriptor for the item type
    cell%    field item-offset \ offset in stack items, 0 for the deepest element
    cell%    field item-first  \ true if this is the first occurence of the item
   end-struct item%
   
   struct%
       cell% 2* field type-c-name
       cell%    field type-stack \ default stack
       cell%    field type-size  \ size of type in stack items
       cell%    field type-fetch \ xt of fetch code generator ( item -- )
       cell%    field type-store \ xt of store code generator ( item -- )
   end-struct type%
   
   struct%
       cell%    field register-number
       cell%    field register-type \ pointer to type
       cell% 2* field register-name \ c name
   end-struct register%
   
   struct%
       cell% 2* field ss-registers  \ addr u; ss-registers[0] is TOS
                                    \ 0 means: use memory
       cell%    field ss-offset     \ stack pointer offset: sp[-offset] is TOS
   end-struct ss% \ stack-state
   
   struct%
       cell%              field state-enabled
       cell%              field state-number
       cell% max-stacks * field state-sss
   end-struct state%
   
   variable next-stack-number 0 next-stack-number !
   create stacks max-stacks cells allot \ array of stacks
   256 constant max-registers
   create registers max-registers cells allot \ array of registers
   variable nregisters 0 nregisters ! \ number of registers
   variable next-state-number 0 next-state-number ! \ next state number
   
   : stack-in-index ( in-size item -- in-index )
       item-offset @ - 1- ;
   
   : inst-in-index ( in-size item -- in-index )
       nip dup item-offset @ swap item-type @ type-size @ + 1- ;
   
   : make-stack ( addr-ptr u1 type "stack-name" -- )
       next-stack-number @ max-stacks < s" too many stacks" ?print-error
       create stack% %allot >r
       r@ stacks next-stack-number @ th !
       next-stack-number @ r@ stack-number !
       1 next-stack-number +!
       r@ stack-type !
       save-mem r@ stack-pointer 2! 
       ['] stack-in-index r@ stack-in-index-xt !
       ['] noop r@ stack-access-transform !
       rdrop ;
   
   : map-stacks { xt -- }
       \ perform xt ( stack -- ) for all stacks
       next-stack-number @ 0 +do
           stacks i th @ xt execute
       loop ;
   
   : map-stacks1 { xt -- }
       \ perform xt ( stack -- ) for all stacks except inst-stream
       next-stack-number @ 1 +do
           stacks i th @ xt execute
       loop ;
   
   \ stack items
   
   : init-item ( addr u addr1 -- )
       \ initialize item at addr1 with name addr u
       \ the stack prefix is removed by the stack-prefix
       dup item% %size erase
       item-name 2! ;
   
   : map-items { addr end xt -- }
       \ perform xt for all items in array addr...end
       end addr ?do
           i xt execute
       item% %size +loop ;
   
 : ` ( -- terminal ) ( use: ` c )  \ types
  ( creates anonymous terminal for the character c )  
  [compile] ascii singleton ['] ?nextchar make-terminal ;  
   
 char a char z ..  char A char Z ..  union char _ singleton union  charclass letter  : print-type-prefix ( type -- )
 char 0 char 9 ..                                        charclass digit      body> >head name>string type ;
 bl singleton                                            charclass blank  
 tab-char singleton                                      charclass tab  
 nl-char singleton eof-char over add-member complement   charclass nonl  
 nl-char singleton eof-char over add-member char : over add-member complement  charclass nocolonnl  
 bl 1+ maxchar ..                                        charclass nowhite  
 char " singleton eof-char over add-member complement    charclass noquote  
 nl-char singleton                                       charclass nl  
 eof-char singleton                                      charclass eof  
   
   \ various variables for storing stuff of one primitive
   
 (( letter (( letter || digit )) **  struct%
 )) <- c-name ( -- )      cell% 2* field prim-name
       cell% 2* field prim-wordset
       cell% 2* field prim-c-name
       cell% 2* field prim-c-name-orig \ for reprocessed prims, the original name
       cell% 2* field prim-doc
       cell% 2* field prim-c-code
       cell% 2* field prim-forth-code
       cell% 2* field prim-stack-string
       cell%    field prim-num            \ ordinal number
       cell%    field prim-items-wordlist \ unique items
       item% max-effect * field prim-effect-in
       item% max-effect * field prim-effect-out
       cell%    field prim-effect-in-end
       cell%    field prim-effect-out-end
       cell% max-stacks * field prim-stacks-in  \ number of in items per stack
       cell% max-stacks * field prim-stacks-out \ number of out items per stack
       cell% max-stacks * field prim-stacks-sync \ sync flag per stack
   end-struct prim%
   
   : make-prim ( -- prim )
       prim% %alloc { p }
       s" " p prim-doc 2! s" " p prim-forth-code 2! s" " p prim-wordset 2!
       p ;
   
   0 value prim     \ in combined prims either combined or a part
   0 value combined \ in combined prims the combined prim
   variable in-part \ true if processing a part
    in-part off
   0 value state-in  \ state on entering prim
   0 value state-out \ state on exiting prim
   0 value state-default  \ canonical state at bb boundaries
   
   : prim-context ( ... p xt -- ... )
       \ execute xt with prim set to p
       prim >r
       swap to prim
       catch
       r> to prim
       throw ;
   
   : prim-c-name-2! ( c-addr u -- )
       2dup prim prim-c-name 2! prim prim-c-name-orig 2! ;
   
   1000 constant max-combined
   create combined-prims max-combined cells allot
   variable num-combined
   variable part-num \ current part number during process-combined
   
   : map-combined { xt -- }
       \ perform xt for all components of the current combined instruction
       num-combined @ 0 +do
           combined-prims i th @ xt execute
       loop ;
   
   table constant combinations
     \ the keys are the sequences of pointers to primitives
   
   create current-depth max-stacks cells allot
   create max-depth     max-stacks cells allot
   create min-depth     max-stacks cells allot
   
   create sp-update-in max-stacks cells allot
   \ where max-depth occured the first time
   create max-depths max-stacks max-combined 1+ * cells allot
   \ maximum depth at start of each part: array[parts] of array[stack]
   create max-back-depths max-stacks max-combined 1+ * cells allot
   \ maximun depth from end of the combination to the start of the each part
   
   : s-c-max-depth ( nstack ncomponent -- addr )
       max-stacks * + cells max-depths + ;
   
   : s-c-max-back-depth ( nstack ncomponent -- addr )
       max-stacks * + cells max-back-depths + ;
   
   wordlist constant primitives
   
   : create-prim ( prim -- )
       dup prim-name 2@ primitives ['] constant insert-wordlist ;
   
   : stack-in ( stack -- addr )
       \ address of number of stack items in effect in
       stack-number @ cells prim prim-stacks-in + ;
   
   : stack-out ( stack -- addr )
       \ address of number of stack items in effect out
       stack-number @ cells prim prim-stacks-out + ;
   
   : stack-prim-stacks-sync ( stack -- addr )
       prim prim-stacks-sync swap stack-number @ th ;
   
   \ global vars
   variable c-line
   2variable c-filename
   variable name-line
   2variable name-filename
   2variable last-name-filename
   Variable function-number 0 function-number !
   Variable function-old 0 function-old !
   : function-diff ( -- )
       ." GROUPADD(" function-number @ function-old @ - 0 .r ." )" cr
       function-number @ function-old ! ;
   : forth-fdiff ( -- )
       function-number @ function-old @ - 0 .r ."  groupadd" cr
       function-number @ function-old ! ;
   
 nowhite ++  \ a few more set ops
 <- name ( -- )  
   
 (( ` \ nonl ** nl  : bit-equivalent ( w1 w2 -- w3 )
 )) <- comment ( -- )   xor invert ;
   
 (( {{ effect-in }} (( {{ start }} c-name {{ end 2 pick item-name 2! item-descr + }} blank ** )) ** {{ effect-in-end ! }}  : complement ( set1 -- set2 )
    ` - ` - blank **   empty ['] bit-equivalent binary-set-operation ;
    {{ effect-out }} (( {{ start }} c-name {{ end 2 pick item-name 2! item-descr + }} blank ** )) ** {{ effect-out-end ! }}  
 )) <- stack-effect ( -- )  
   
 (( {{ s" " doc 2! s" " forth-code 2! }}  \ forward declaration for inst-stream (breaks cycle in definitions)
    (( comment || nl )) **  defer inst-stream-f ( -- stack )
    (( {{ start }} name {{ end 2dup forth-name 2! c-name 2! }}  tab ++  
       {{ start }} stack-effect {{ end stack-string 2! }} tab ++  
         {{ start }} name {{ end wordset 2! }} tab **  
         (( {{ start }}  c-name {{ end c-name 2! }} )) ??  nl  
    ))  
    (( ` " ` "  {{ start }} (( noquote ++ ` " )) ++ {{ end 1- doc 2! }} ` " nl )) ??  
    {{ start }} (( nocolonnl nonl **  nl )) ** {{ end c-code 2! }}  
    (( ` :  nl  
       {{ start }} (( nonl ++  nl )) ++ {{ end forth-code 2! }}  
    )) ??  
    (( nl || eof ))  
 )) <- primitive ( -- )  
   
 (( (( primitive {{ printprim }} )) **  eof ))  \ stack access stuff
 parser primitives2something  
 warnings @ [IF]  
 .( parser generated ok ) cr  
 [THEN]  
   
 : primfilter ( file-id xt -- )  : normal-stack-access0 { n stack -- }
 \ fileid is for the input file, xt ( -- ) is for the output word      \ n has the ss-offset already applied (see ...-access1)
  output !      n stack stack-access-transform @ execute ." [" 0 .r ." ]" ;
  here input !  
  here swap read-whole-file  : state-ss { stack state -- ss }
  dup endinput !      state state-sss stack stack-number @ th @ ;
  here - allot  
  align  : stack-reg { n stack state -- reg }
  primitives2something ;      \ n is the index (TOS=0); reg is 0 if the access is to memory
       stack state state-ss ss-registers 2@ n u> if ( addr ) \ in ss-registers?
           n th @
       else
           drop 0
       endif ;
   
   : .reg ( reg -- )
       register-name 2@ type ;
   
   : stack-offset ( stack state -- n )
       \ offset for stack in state
       state-ss ss-offset @ ;
   
   : normal-stack-access1 { n stack state -- }
       n stack state stack-reg ?dup-if
           .reg exit
       endif
       stack stack-pointer 2@ type
       n stack state stack-offset - stack normal-stack-access0 ;
   
 \ types  : normal-stack-access ( n stack state -- )
       over inst-stream-f = if
           ." IMM_ARG(" normal-stack-access1 ." ," immarg ? ." )"
           1 immarg +!
       else
           normal-stack-access1
       endif ;
   
   : stack-depth { stack -- n }
       current-depth stack stack-number @ th @ ;
   
   : part-stack-access { n stack -- }
       \ print _<stack><x>, x=inst-stream? n : maxdepth-currentdepth-n-1
       ." _" stack stack-pointer 2@ type
       stack stack-number @ { stack# }
       stack stack-depth n + { access-depth }
       stack inst-stream-f = if
           access-depth
       else
           combined prim-stacks-in stack# th @
           assert( dup max-depth stack# th @ = )
           access-depth - 1-
       endif
       0 .r ;
   
 struct  : part-stack-read { n stack -- }
  2 cells field type-c-name      stack stack-depth n + ( ndepth )
  cell const-field type-d-size      stack stack-number @ part-num @ s-c-max-depth @
  cell const-field type-f-size  \    max-depth stack stack-number @ th @ ( ndepth nmaxdepth )
  cell const-field type-fetch-handler      over <= if ( ndepth ) \ load from memory
  cell const-field type-store-handler          stack state-in normal-stack-access
 constant type-description      else
           drop n stack part-stack-access
 : data-stack-access ( n1 n2 n3 -- )      endif ;
 \ n1 is the offset of the accessed item, n2, n3 are effect-*-size  
  drop swap - 1- dup  : stack-diff ( stack -- n )
  if      \ in-out
    ." sp[" 0 .r ." ]"      dup stack-in @ swap stack-out @ - ;
  else  
    drop ." TOS"  : part-stack-write { n stack -- }
  endif ;      stack stack-depth n +
       stack stack-number @ part-num @ s-c-max-back-depth @
       over <= if ( ndepth )
           stack combined ['] stack-diff prim-context -
           stack state-out normal-stack-access
       else
           drop n stack part-stack-access
       endif ;
   
   : stack-read ( n stack -- )
       \ print a stack access at index n of stack
       in-part @ if
           part-stack-read
       else
           state-in normal-stack-access
       endif ;
   
   : stack-write ( n stack -- )
       \ print a stack access at index n of stack
       in-part @ if
           part-stack-write
       else
           state-out normal-stack-access
       endif ;
   
   : item-in-index { item -- n }
       \ n is the index of item (in the in-effect)
       item item-stack @ dup >r stack-in @ ( in-size r:stack )
       item r> stack-in-index-xt @ execute ;
   
 : fp-stack-access ( n1 n2 n3 -- )  : item-stack-type-name ( item -- addr u )
 \ n1 is the offset of the accessed item, n2, n3 are effect-*-size      item-stack @ stack-type @ type-c-name 2@ ;
  nip swap - 1- dup  
  if  
    ." fp[" 0 .r ." ]"  
  else  
    drop ." FTOS"  
  endif ;  
   
 : fetch-single ( item -- )  : fetch-single ( item -- )
  >r      \ fetch a single stack item from its stack
  r@ item-name 2@ type      >r
  ."  = ("       ." vm_" r@ item-stack-type-name type
  r@ item-type @ type-c-name 2@ type ." ) "      ." 2" r@ item-type @ print-type-prefix ." ("
  r@ item-d-offset @ effect-in-size 2@ data-stack-access ." ;" cr      r@ item-in-index r@ item-stack @ stack-read ." ,"
  rdrop ;       r@ item-name 2@ type
       ." );" cr
       rdrop ; 
   
 : fetch-double ( item -- )  : fetch-double ( item -- )
  >r      \ fetch a double stack item from its stack
  r@ item-name 2@ type       >r
  ." = ({Double_Store _d; _d.cells.low = "      ." vm_two"
  r@ item-d-offset @ dup    effect-in-size 2@ data-stack-access      r@ item-stack-type-name type ." 2"
  ." ; _d.cells.high = " 1+ effect-in-size 2@ data-stack-access      r@ item-type @ print-type-prefix ." ("
  ." ; _d.dcell;});" cr      r@ item-in-index r@ item-stack @ 2dup stack-read
  rdrop ;      ." , "                      -1 under+ stack-read
       ." , " r@ item-name 2@ type
 : fetch-float ( item -- )      ." )" cr
  >r      rdrop ;
  r@ item-name 2@ type  
  ."  = "  : same-as-in? ( item -- f )
  \ ." (" r@ item-type @ type-c-name 2@ type ." ) "      \ f is true iff the offset and stack of item is the same as on input
  r@ item-f-offset @ effect-in-size 2@ fp-stack-access ." ;" cr      >r
  rdrop ;      r@ item-stack @ stack-prim-stacks-sync @ if
           rdrop false exit
 : d-same-as-in? ( item -- f )      endif
 \ f is true iff the offset of item is the same as on input      r@ item-first @ if
  >r          rdrop false exit
  r@ item-name 2@ items @ search-wordlist 0=      endif
  abort" bug"      r@ item-name 2@ prim prim-items-wordlist @ search-wordlist 0= abort" bug"
  execute @      execute @
  dup r@ =      dup r@ =
  if \ item first appeared in output      if \ item first appeared in output
    drop false          drop false
  else      else
    item-d-offset @ r@ item-d-offset @ =          dup  item-stack  @ r@ item-stack  @ = 
  endif          swap item-offset @ r@ item-offset @ = and
  rdrop ;      endif
       rdrop ;
   
 : is-in-tos? ( item -- f )  : item-out-index ( item -- n )
 \ true if item has the same offset as the input TOS      \ n is the index of item (in the out-effect)
  item-d-offset @ 1+ effect-in-size 2@ drop = ;      >r r@ item-stack @ stack-out @ r> item-offset @ - 1- ;
   
 : really-store-single ( item -- )  : really-store-single ( item -- )
  >r      >r
  r@ item-d-offset @ effect-out-size 2@ data-stack-access ."  = (Cell)"      ." vm_"
  r@ item-name 2@ type ." ;"      r@ item-type @ print-type-prefix ." 2"
  rdrop ;      r@ item-stack-type-name type ." ("
       r@ item-name 2@ type ." ,"
 : store-single ( item -- )      r@ item-out-index r@ item-stack @ stack-write ." );"
  >r      rdrop ;
  r@ d-same-as-in?  
  if  : store-single { item -- }
    r@ is-in-tos?      item item-stack @ { stack }
    if      store-optimization @ in-part @ 0= and item same-as-in? and
      ." IF_TOS(" r@ really-store-single ." );" cr      item item-in-index  stack state-in  stack-reg       \  in reg/mem
    endif      item item-out-index stack state-out stack-reg = and \ out reg/mem
  else      0= if
    r@ really-store-single cr          item really-store-single cr
  endif      endif ;
  rdrop ;  
   
 : store-double ( item -- )  : store-double ( item -- )
 \ !! store optimization is not performed, because it is not yet needed  \ !! store optimization is not performed, because it is not yet needed
  >r   >r
  ." {Double_Store _d; _d.dcell = " r@ item-name 2@ type ." ; "   ." vm_"
  r@ item-d-offset @ dup    effect-out-size 2@ data-stack-access    r@ item-type @ print-type-prefix ." 2two"
  ."  = _d.cells.low; " 1+ effect-out-size 2@ data-stack-access   r@ item-stack-type-name type ." ("
  ." = _d.cells.high;}" cr   r@ item-name 2@ type ." , "
  rdrop ;   r@ item-out-index r@ item-stack @ 2dup stack-write
    ." , "                       -1 under+ stack-write
 : f-same-as-in? ( item -- f )   ." )" cr
 \ f is true iff the offset of item is the same as on input  
  >r  
  r@ item-name 2@ items @ search-wordlist 0=  
  abort" bug"  
  execute @  
  dup r@ =  
  if \ item first appeared in output  
    drop false  
  else  
    item-f-offset @ r@ item-f-offset @ =  
  endif  
  rdrop ;  
   
 : is-in-ftos? ( item -- f )  
 \ true if item has the same offset as the input TOS  
  item-f-offset @ 1+ effect-in-size 2@ nip = ;  
   
 : really-store-float ( item -- )  
  >r  
  r@ item-f-offset @ effect-out-size 2@ fp-stack-access ."  = "  
  r@ item-name 2@ type ." ;"  
  rdrop ;  
   
 : store-float ( item -- )  
  >r  
  r@ f-same-as-in?  
  if  
    r@ is-in-ftos?  
    if  
      ." IF_FTOS(" r@ really-store-float ." );" cr  
    endif  
  else  
    r@ really-store-float cr  
  endif  
  rdrop ;   rdrop ;
    
 : single-type ( -- xt1 xt2 n1 n2 )  
  ['] fetch-single ['] store-single 1 0 ;  
   
 : double-type ( -- xt1 xt2 n1 n2 )  : single ( -- xt1 xt2 n )
  ['] fetch-double ['] store-double 2 0 ;      ['] fetch-single ['] store-single 1 ;
   
 : float-type ( -- xt1 xt2 n1 n2 )  : double ( -- xt1 xt2 n )
  ['] fetch-float ['] store-float 0 1 ;      ['] fetch-double ['] store-double 2 ;
   
 : s, ( addr u -- )  : s, ( addr u -- )
 \ allocate a string  \ allocate a string
  here swap dup allot move ;   here swap dup allot move ;
   
 : starts-with ( addr u xt1 xt2 n1 n2 "prefix" -- )  wordlist constant prefixes
 \ describes a type  
 \ addr u specifies the C type name  
 \ n1 is the size of the type on the data stack  
 \ n2 is the size of the type on the FP stack  
 \ stack effect entries of the type start with prefix  
  >r >r >r >r  
  dup >r here >r s,  
  create  
  r> r> 2,  
  r> r> r> , r> , swap , , ;  
   
 wordlist constant types  
 get-current  
 types set-current  
   
 s" Bool"        single-type starts-with f  
 s" Char"        single-type starts-with c  
 s" Cell"        single-type starts-with n  
 s" Cell"        single-type starts-with w  
 s" UCell"       single-type starts-with u  
 s" DCell"       double-type starts-with d  
 s" UDCell"      double-type starts-with ud  
 s" Float"       float-type  starts-with r  
 s" Cell *"      single-type starts-with a_  
 s" Char *"      single-type starts-with c_  
 s" Float *"     single-type starts-with f_  
 s" DFloat *"    single-type starts-with df_  
 s" SFloat *"    single-type starts-with sf_  
 s" Xt"          single-type starts-with xt  
 s" WID"         single-type starts-with wid  
 s" F83Name *"   single-type starts-with f83name  
   
 set-current  
   
 : get-type ( addr1 u1 -- type-descr )  
 \ get the type of the name in addr1 u1  
 \ type-descr is a pointer to a type-descriptor  
  0 swap ?do  
    dup i types search-wordlist  
    if \ ok, we have the type ( addr1 xt )  
      execute nip  
      UNLOOP EXIT  
    endif  
  -1 s+loop  
  \ we did not find a type, abort  
  true abort" unknown type prefix" ;  
   
 : declare ( addr "name" -- )  : declare ( addr "name" -- )
 \ remember that there is a stack item at addr called name  \ remember that there is a stack item at addr called name
  create , ;   create , ;
   
   : !default ( w addr -- )
       dup @ if
           2drop \ leave nonzero alone
       else
           !
       endif ;
   
   : create-type { addr u xt1 xt2 n stack -- } ( "prefix" -- )
       \ describes a type
       \ addr u specifies the C type name
       \ stack effect entries of the type start with prefix
       create type% %allot >r
       addr u save-mem r@ type-c-name 2!
       xt1   r@ type-fetch !
       xt2   r@ type-store !
       n     r@ type-size !
       stack r@ type-stack !
       rdrop ;
   
   : type-prefix ( addr u xt1 xt2 n stack "prefix" -- )
       get-current >r prefixes set-current
       create-type r> set-current
   does> ( item -- )
       \ initialize item
       { item typ }
       typ item item-type !
       typ type-stack @ item item-stack !default
       item item-name 2@ prim prim-items-wordlist @ search-wordlist 0= if
           item item-name 2@ nextname item declare
           item item-first on
           \ typ type-c-name 2@ type space type  ." ;" cr
       else
           drop
           item item-first off
       endif ;
   
   : execute-prefix ( item addr1 u1 -- )
       \ execute the word ( item -- ) associated with the longest prefix
       \ of addr1 u1
       0 swap ?do
           dup i prefixes search-wordlist
           if \ ok, we have the type ( item addr1 xt )
               nip execute
               UNLOOP EXIT
           endif
           -1 s+loop
           \ we did not find a type, abort
           abort
       false s" unknown prefix" ?print-error ;
   
 : declaration ( item -- )  : declaration ( item -- )
  dup item-name 2@ items @ search-wordlist      dup item-name 2@ execute-prefix ;
  if \ already declared ( item xt )  
    execute @ item-type @ swap item-type !  
  else ( addr )  
    dup item-name 2@ nextname dup declare ( addr )  
    dup >r item-name 2@ 2dup get-type ( addr1 u type-descr )  
    dup r> item-type ! ( addr1 u type-descr )  
    type-c-name 2@ type space type ." ;" cr  
  endif ;  
   
 : declaration-list ( addr1 addr2 -- )  : declaration-list ( addr1 addr2 -- )
  swap ?do      ['] declaration map-items ;
   i declaration  
  item-descr +loop ;  
   
 : fetch ( addr -- )  
  dup item-type @ type-fetch-handler execute ;  
   
 : declarations ( -- )  : declarations ( -- )
  wordlist dup items ! set-current   wordlist dup prim prim-items-wordlist ! set-current
  effect-in effect-in-end @ declaration-list   prim prim-effect-in prim prim-effect-in-end @ declaration-list
  effect-out effect-out-end @ declaration-list ;   prim prim-effect-out prim prim-effect-out-end @ declaration-list ;
   
   : print-declaration { item -- }
       item item-first @ if
           item item-type @ type-c-name 2@ type space
           item item-name 2@ type ." ;" cr
       endif ;
   
   : print-declarations ( -- )
       prim prim-effect-in  prim prim-effect-in-end  @ ['] print-declaration map-items
       prim prim-effect-out prim prim-effect-out-end @ ['] print-declaration map-items ;
       
   : stack-prefix ( stack "prefix" -- )
       get-current >r prefixes set-current
       name tuck nextname create ( stack length ) 2,
       r> set-current
   does> ( item -- )
       2@ { item stack prefix-length }
       item item-name 2@ prefix-length /string item item-name 2!
       stack item item-stack !
       item declaration ;
   
   : set-prim-stacks-sync ( stack -- )
       stack-prim-stacks-sync on ;
   
   : clear-prim-stacks-sync ( stack -- )
       stack-prim-stacks-sync off ;
   
   
   get-current prefixes set-current
   : ... ( item -- )
       \ this "prefix" ensures that the appropriate stack is synced with memory
       dup item-name 2@ s" ..." str= 0= abort" '...' must end the item name"
       item-stack @ dup if
           set-prim-stacks-sync
       else \ prefixless "..." syncs all stacks
           drop ['] set-prim-stacks-sync map-stacks1
       endif ;
   set-current
   
   create ...-item ( -- addr ) \ just used for letting stack-prefixes work on it
   item% %allot drop           \ stores the stack temporarily until used by ...
   
   : init-item1 ( addr1 addr u -- addr2 )
       \ initialize item at addr1 with name addr u, next item is at addr2
       \ !! make sure that any mention of "..." is only stack-prefixed
       2dup s" ..." search nip nip if ( addr1 addr u )
           0 ...-item item-stack ! \ initialize to prefixless
           2dup ...-item item-name 2!
           ...-item rot rot execute-prefix ( addr1 )
       else
           2 pick init-item item% %size +
       endif ;
   
   \ types pointed to by stacks for use in combined prims
   \ !! output-c-combined shouldn't use these names!
   : stack-type-name ( addr u "name" -- )
       single 0 create-type ;
   
   wordlist constant type-names \ this is here just to meet the requirement
                       \ that a type be a word; it is never used for lookup
   
   : define-type ( addr u -- xt )
       \ define single type with name addr u, without stack
       get-current type-names set-current >r
       2dup nextname stack-type-name
       r> set-current
       latestxt ;
   
   : stack ( "name" "stack-pointer" "type" -- )
       \ define stack
       name { d: stack-name }
       name { d: stack-pointer }
       name { d: stack-type }
       stack-type define-type
       stack-pointer rot >body stack-name nextname make-stack ;
   
   stack inst-stream IP Cell
   ' inst-in-index inst-stream stack-in-index-xt !
   ' inst-stream <is> inst-stream-f
   \ !! initialize stack-in and stack-out
   
   \ registers
   
   : make-register ( type addr u -- )
       \ define register with type TYPE and name ADDR U.
       nregisters @ max-registers < s" too many registers" ?print-error
       2dup nextname create register% %allot >r
       r@ register-name 2!
       r@ register-type !
       nregisters @ r@ register-number !
       1 nregisters +!
       rdrop ;
   
   : register ( "name" "type" -- )
       \ define register
       name { d: reg-name }
       name { d: reg-type }
       reg-type define-type >body
       reg-name make-register ;
   
   \ stack-states
   
   : stack-state ( a-addr u uoffset "name" -- )
       create ss% %allot >r
       r@ ss-offset !
       r@ ss-registers 2!
       rdrop ;
   
   0 0 0 stack-state default-ss
   
   \ state
   
   : state ( "name" -- )
       \ create a state initialized with default-sss
       create state% %allot { s }
       s state-enabled on
       next-state-number @ s state-number ! 1 next-state-number +!
       max-stacks 0 ?do
           default-ss s state-sss i th !
       loop ;
   
   : state-disable ( state -- )
       state-enabled off ;
   
   : state-enabled? ( state -- f )
       state-enabled @ ;
   
   : .state ( state -- )
       0 >body - >name .name ;
   
   : set-ss ( ss stack state -- )
       state-sss swap stack-number @ th ! ;
   
 \ offset computation  \ offset computation
 \ the leftmost (i.e. deepest) item has offset 0  \ the leftmost (i.e. deepest) item has offset 0
 \ the rightmost item has the highest offset  \ the rightmost item has the highest offset
   
 : compute-offset ( n1 n2 item -- n3 n4 )  : compute-offset { item xt -- }
 \ n1, n3 are data-stack-offsets      \ xt specifies in/out; update stack-in/out and set item-offset
 \ n2, n4 are the fp-stack-offsets      item item-type @ type-size @
  >r      item item-stack @ xt execute dup @ >r +!
  swap dup r@ item-d-offset !      r> item item-offset ! ;
  r@ item-type @ type-d-size +  
  swap dup r@ item-f-offset !  
  r@ item-type @ type-f-size +  
  rdrop ;  
   
 : compute-list ( addr1 addr2 -- n1 n2 )  : compute-offset-in ( addr1 addr2 -- )
 \ n1, n2 are the final offsets      ['] stack-in compute-offset ;
  0 0 2swap swap ?do  
   i compute-offset  : compute-offset-out ( addr1 addr2 -- )
  item-descr +loop ;      ['] stack-out compute-offset ;
   
 : compute-offsets ( -- )  : compute-offsets ( -- )
  effect-in effect-in-end @ compute-list effect-in-size 2!      prim prim-stacks-in  max-stacks cells erase
  effect-out effect-out-end @ compute-list effect-out-size 2! ;      prim prim-stacks-out max-stacks cells erase
       prim prim-effect-in  prim prim-effect-in-end  @ ['] compute-offset-in  map-items
       prim prim-effect-out prim prim-effect-out-end @ ['] compute-offset-out map-items
       inst-stream stack-out @ 0= s" # can only be on the input side" ?print-error ;
   
   : init-simple { prim -- }
       \ much of the initialization is elsewhere
       ['] clear-prim-stacks-sync map-stacks ;
   
   : process-simple ( -- )
       prim prim { W^ key } key cell
       combinations ['] constant insert-wordlist
       declarations compute-offsets
       output @ execute ;
   
   : stack-state-items ( stack state -- n )
       state-ss ss-registers 2@ nip ;
   
   : unused-stack-items { stack -- n-in n-out }
       \ n-in  are the stack items in state-in  not used    by prim
       \ n-out are the stack items in state-out not written by prim
       stack state-in  stack-state-items stack stack-in  @ - 0 max
       stack state-out stack-state-items stack stack-out @ - 0 max ;
   
   : spill-stack-items { stack -- u }
       \ there are u items to spill in stack
       stack unused-stack-items
       stack stack-prim-stacks-sync @ if
           drop 0
       endif
       swap - ;
   
 : flush-tos ( -- )  : spill-stack { stack -- }
  effect-in-size 2@ effect-out-size 2@      \ spill regs of state-in that are not used by prim and are not in state-out
  0<> rot 0= and      stack state-in stack-offset { offset }
  if      stack state-in stack-state-items ( items )
    ." IF_FTOS(fp[0] = FTOS);" cr      dup stack spill-stack-items + +do
  endif          \ loop through the bottom items
  0<> swap 0= and          stack stack-pointer 2@ type
  if          i offset - stack normal-stack-access0 ."  = "
    ." IF_TOS(sp[0] = TOS);" cr          i stack state-in normal-stack-access1 ." ;" cr
  endif ;      loop ;
   
   : spill-state ( -- )
       ['] spill-stack map-stacks1 ;
   
   : fill-stack-items { stack -- u }
       \ there are u items to fill in stack
       stack unused-stack-items
       stack stack-prim-stacks-sync @ if
           swap drop 0 swap
       endif
       - ;
   
 : fill-tos ( -- )  : fill-stack { stack -- }
  effect-in-size 2@ effect-out-size 2@      stack state-out stack-offset { offset }
  0= rot 0<> and      stack state-out stack-state-items ( items )
  if      dup stack fill-stack-items + +do
    ." IF_FTOS(FTOS = fp[0]);" cr          \ loop through the bottom items
  endif          i stack state-out normal-stack-access1 ."  = "
  0= swap 0<> and          stack stack-pointer 2@ type
  if          i offset - stack normal-stack-access0 ." ;" cr
    ." IF_TOS(TOS = sp[0]);" cr      loop ;
  endif ;  
   : fill-state ( -- )
       \ !! inst-stream for prefetching?
       ['] fill-stack map-stacks1 ;
   
   : fetch ( addr -- )
       dup item-type @ type-fetch @ execute ;
   
 : fetches ( -- )  : fetches ( -- )
  effect-in-end @ effect-in ?do      prim prim-effect-in prim prim-effect-in-end @ ['] fetch map-items ;
    i fetch  
  item-descr +loop ;   : reg-reg-move ( reg-from reg-to -- )
       2dup = if
           2drop
       else
           .reg ."  = " .reg ." ;" cr
       endif ;
   
   : stack-bottom-reg { n stack state -- reg }
       stack state stack-state-items n - 1- stack state stack-reg ;
   
   : stack-moves { stack -- }
       \ generate moves between registers in state-in/state-out that are
       \ not spilled or consumed/produced by prim.
       \ !! this works only for a simple stack cache, not e.g., for
       \ rotating stack caches, or registers shared between stacks (the
       \ latter would also require a change in interface)
       \ !! maybe place this after NEXT_P1?
       stack unused-stack-items 2dup < if ( n-in n-out )
           \ move registers from 0..n_in-1 to n_out-n_in..n_out-1
           over - { diff } ( n-in )
           -1 swap 1- -do
               i stack state-in stack-bottom-reg ( reg-from )
               i diff + stack state-out stack-bottom-reg reg-reg-move
           1 -loop
       else
           \ move registers from n_in-n_out..n_in-1 to 0..n_out-1
           swap over - { diff } ( n-out )
           0 +do
               i diff + stack state-in stack-bottom-reg ( reg-from )
               i stack state-out stack-bottom-reg reg-reg-move
           loop
       endif ;
   
   : stack-update-transform ( n1 stack -- n2 )
       \ n2 is the number by which the stack pointer should be
       \ incremented to pop n1 items
       stack-access-transform @ dup >r execute
       0 r> execute - ;
   
   : update-stack-pointer { stack n -- }
       n if \ this check is not necessary, gcc would do this for us
           stack inst-stream = if
               ." INC_IP(" n 0 .r ." );" cr
           else
               stack stack-pointer 2@ type ."  += "
               n stack stack-update-transform 0 .r ." ;" cr
           endif
       endif ;
   
   : stack-pointer-update { stack -- }
       \ and moves
       \ stacks grow downwards
       stack stack-prim-stacks-sync @ if
           stack stack-in @
           stack state-in  stack-offset -
           stack swap update-stack-pointer
       else
           stack stack-diff ( in-out )
           stack state-in  stack-offset -
           stack state-out stack-offset + ( [in-in_offset]-[out-out_offset] )
           stack swap update-stack-pointer
           stack stack-moves
       endif ;
   
 : stack-pointer-updates ( -- )  : stack-pointer-updates ( -- )
 \ we need not check if an update is a noop; gcc does this for us      ['] stack-pointer-update map-stacks ;
  effect-in-size 2@  
  effect-out-size 2@  : stack-pointer-update2 { stack -- }
  rot swap - ( d-in d-out f-diff )      stack stack-prim-stacks-sync @ if
  rot rot - ( f-diff d-diff )          stack state-out stack-offset
  ?dup IF  ." sp += " 0 .r ." ;" cr  THEN          stack stack-out @ -
  ?dup IF  ." fp += " 0 .r ." ;" cr  THEN ;          stack swap update-stack-pointer
       endif ;
   
   : stack-pointer-updates2 ( -- )
       \ update stack pointers after C code, where necessary
       ['] stack-pointer-update2 map-stacks ;
   
 : store ( item -- )  : store ( item -- )
 \ f is true if the item should be stored  \ f is true if the item should be stored
 \ f is false if the store is probably not necessary  \ f is false if the store is probably not necessary
  dup item-type @ type-store-handler execute ;   dup item-type @ type-store @ execute ;
   
 : stores ( -- )  : stores ( -- )
  effect-out-end @ effect-out ?do      prim prim-effect-out prim prim-effect-out-end @ ['] store map-items ;
    i store  
  item-descr +loop ;   : print-debug-arg { item -- }
       ." fputs(" quote space item item-name 2@ type ." =" quote ." , vm_out); "
 : .stack-list ( start end -- )      ." printarg_" item item-type @ print-type-prefix
  swap ?do      ." (" item item-name 2@ type ." );" cr ;
    i item-name 2@ type space      
  item-descr +loop ;   : print-debug-args ( -- )
       ." #ifdef VM_DEBUG" cr
       ." if (vm_debug) {" cr
       prim prim-effect-in prim prim-effect-in-end @ ['] print-debug-arg map-items
   \    ." fputc('\n', vm_out);" cr
       ." }" cr
       ." #endif" cr ;
   
   : print-debug-result { item -- }
       item item-first @ if
           item print-debug-arg
       endif ;
   
   : print-debug-results ( -- )
       cr
       ." #ifdef VM_DEBUG" cr
       ." if (vm_debug) {" cr
       ." fputs(" quote ."  -- " quote ." , vm_out); "
       prim prim-effect-out prim prim-effect-out-end @ ['] print-debug-result map-items
       ." fputc('\n', vm_out);" cr
       ." }" cr
       ." #endif" cr ;
   
   : output-super-end ( -- )
       prim prim-c-code 2@ s" SET_IP" search if
           ." SUPER_END;" cr
       endif
       2drop ;
   
   
   defer output-nextp0
   :noname ( -- )
       ." NEXT_P0;" cr ;
   is output-nextp0
   
   defer output-nextp1
   :noname ( -- )
       ." NEXT_P1;" cr ;
   is output-nextp1
   
   : output-nextp2 ( -- )
       ." NEXT_P2;" cr ;
   
   variable tail-nextp2 \ xt to execute for printing NEXT_P2 in INST_TAIL
   ' output-nextp2 tail-nextp2 !
   
   : output-label2 ( -- )
       ." LABEL2(" prim prim-c-name 2@ type ." )" cr
       ." NEXT_P1_5;" cr
       ." LABEL3(" prim prim-c-name 2@ type ." )" cr
       ." DO_GOTO;" cr ;
   
   : output-c-tail1 { xt -- }
       \ the final part of the generated C code, with xt printing LABEL2 or not.
       output-super-end
       print-debug-results
       output-nextp1
       stack-pointer-updates2
       stores
       fill-state 
       xt execute ;
   
   : output-c-vm-jump-tail ( -- )
       \ !! this functionality not yet implemented for superinstructions
       output-super-end
       print-debug-results
       stores
       fill-state 
       ." LABEL2(" prim prim-c-name 2@ type ." )" cr
       ." LABEL3(" prim prim-c-name 2@ type ." )" cr
       ." DO_GOTO;" cr ;
   
   : output-c-tail1-no-stores { xt -- }
       \ the final part of the generated C code for combinations
       output-super-end
       output-nextp1
       fill-state 
       xt execute ;
   
   : output-c-tail ( -- )
       tail-nextp2 @ output-c-tail1 ;
   
   : output-c-tail2 ( -- )
       prim prim-c-code 2@ s" VM_JUMP(" search nip nip if
           output-c-vm-jump-tail
       else
           ['] output-label2 output-c-tail1
       endif ;
   
   : output-c-tail-no-stores ( -- )
       tail-nextp2 @ output-c-tail1-no-stores ;
   
   : output-c-tail2-no-stores ( -- )
       ['] output-label2 output-c-tail1-no-stores ;
   
   : type-c-code ( c-addr u xt -- )
       \ like TYPE, but replaces "INST_TAIL;" with tail code produced by xt
       { xt }
       ." {" cr
       ." #line " c-line @ . quote c-filename 2@ type quote cr
       begin ( c-addr1 u1 )
           2dup s" INST_TAIL;" search
       while ( c-addr1 u1 c-addr3 u3 )
           2dup 2>r drop nip over - type
           xt execute
           2r> 10 /string
           \ !! resync #line missing
       repeat
       2drop type
       ." #line " out-nls @ 2 + . quote out-filename 2@ type quote cr
       ." }" cr ;
   
   : print-entry ( -- )
       ." LABEL(" prim prim-c-name 2@ type ." )" ;
   
   : prim-type ( addr u -- )
       \ print out a primitive, but avoid "*/"
       2dup s" */" search  nip nip  IF
           bounds ?DO  I c@ dup '* = IF  drop 'x  THEN  emit  LOOP
       ELSE  type  THEN ;
   
 : output-c ( -- )  : output-c ( -- )
  ." I_" c-name 2@ type ." :     /* " forth-name 2@ type ."  ( " stack-string 2@ type ."  ) */" cr      print-entry ."  /* " prim prim-name 2@ prim-type
  ." /* " doc 2@ type ."  */" cr      ."  ( " prim prim-stack-string 2@ type ." ) "
  ." NAME(" [char] " emit forth-name 2@ type [char] " emit ." )" cr \ debugging      state-in .state ." -- " state-out .state ."  */" cr
  ." {" cr      ." /* " prim prim-doc 2@ type ."  */" cr
  ." DEF_CA" cr      ." NAME(" quote prim prim-name 2@ type quote ." )" cr \ debugging
  declarations      ." {" cr
  compute-offsets \ for everything else      ." DEF_CA" cr
  ." NEXT_P0;" cr      print-declarations
  flush-tos      output-nextp0
  fetches      spill-state
  stack-pointer-updates      fetches
  ." {" cr      print-debug-args
  c-code 2@ type      stack-pointer-updates
  ." }" cr      prim prim-c-code 2@ ['] output-c-tail type-c-code
  ." NEXT_P1;" cr      output-c-tail2
  stores      ." }" cr
  fill-tos      cr
  ." NEXT_P2;" cr  
  ." }" cr  
  cr  
 ;  ;
   
 : output-label ( -- )  : disasm-arg { item -- }
  ." &&I_" c-name 2@ type ." ," cr ;      item item-stack @ inst-stream = if
           ." {" cr
           item print-declaration
           item fetch
           item print-debug-arg
           ." }" cr
       endif ;
   
   : disasm-args ( -- )
       prim prim-effect-in prim prim-effect-in-end @ ['] disasm-arg map-items ;
   
   : output-disasm ( -- )
       \ generate code for disassembling VM instructions
       ." if (VM_IS_INST(*ip, " function-number @ 0 .r ." )) {" cr
       ."   fputs(" quote prim prim-name 2@ type quote ." , vm_out);" cr
       disasm-args
       ."   ip += " inst-stream stack-in @ 1+ 0 .r ." ;" cr
       ."   goto _endif_;" cr
       ." }" cr ;
   
   : output-profile ( -- )
       \ generate code for postprocessing the VM block profile stuff
       ." if (VM_IS_INST(*ip, " function-number @ 0 .r ." )) {" cr
       ."   add_inst(b, " quote prim prim-name 2@ type quote ." );" cr
       ."   ip += " inst-stream stack-in @ 1+ 0 .r ." ;" cr
       prim prim-c-code 2@  s" SET_IP"    search nip nip
       prim prim-c-code 2@  s" SUPER_END" search nip nip or if
           ."   return;" cr
       else
           ."   goto _endif_;" cr
       endif
       ." }" cr ;
   
   : output-profile-part ( p )
       ."   add_inst(b, " quote
       prim-name 2@ type
       quote ." );" cr ;
       
   : output-profile-combined ( -- )
       \ generate code for postprocessing the VM block profile stuff
       ." if (VM_IS_INST(*ip, " function-number @ 0 .r ." )) {" cr
       ['] output-profile-part map-combined
       ."   ip += " inst-stream stack-in @ 1+ 0 .r ." ;" cr
       combined-prims num-combined @ 1- th @ prim-c-code 2@  s" SET_IP"    search nip nip
       combined-prims num-combined @ 1- th @ prim-c-code 2@  s" SUPER_END" search nip nip or if
           ."   return;" cr
       else
           ."   goto _endif_;" cr
       endif
       ." }" cr ;
   
 : output-alias ( -- )  : prim-branch? { prim -- f }
  primitive-number @ . ." alias " forth-name 2@ type cr      \ true if prim is a branch or super-end
  -1 primitive-number +! ;      prim prim-c-code 2@  s" SET_IP" search nip nip 0<> ;
   
 : output-forth ( -- )  : output-superend ( -- )
  forth-code @ 0=      \ output flag specifying whether the current word ends a dynamic superinst
  IF    output-alias      prim prim-branch?
  ELSE  ." : " forth-name 2@ type ."   ( "      prim prim-c-code 2@  s" SUPER_END" search nip nip 0<> or
        effect-in effect-in-end @ .stack-list ." -- "      prim prim-c-code 2@  s" SUPER_CONTINUE" search nip nip 0= and
        effect-out effect-out-end @ .stack-list ." )" cr      negate 0 .r ." , /* " prim prim-name 2@ prim-type ."  */" cr ;
        forth-code 2@ type cr  
        -1 primitive-number +!  : gen-arg-parm { item -- }
  THEN ;      item item-stack @ inst-stream = if
           ." , " item item-type @ type-c-name 2@ type space
           item item-name 2@ type
       endif ;
   
   : gen-args-parm ( -- )
       prim prim-effect-in prim prim-effect-in-end @ ['] gen-arg-parm map-items ;
   
   : gen-arg-gen { item -- }
       item item-stack @ inst-stream = if
           ."   genarg_" item item-type @ print-type-prefix
           ." (ctp, " item item-name 2@ type ." );" cr
       endif ;
   
   : gen-args-gen ( -- )
       prim prim-effect-in prim prim-effect-in-end @ ['] gen-arg-gen map-items ;
   
   : output-gen ( -- )
       \ generate C code for generating VM instructions
       ." void gen_" prim prim-c-name 2@ type ." ("
       threaded-code-pointer-type 2@ type ." ctp" gen-args-parm ." )" cr
       ." {" cr
       ."   gen_inst(ctp, " function-number @ 0 .r ." );" cr
       gen-args-gen
       ." }" cr ;
   
   : stack-used? { stack -- f }
       stack stack-in @ stack stack-out @ or 0<> ;
   
   : output-funclabel ( -- )
     ." &I_" prim prim-c-name 2@ type ." ," cr ;
   
   : output-forthname ( -- )
     '" emit prim prim-name 2@ type '" emit ." ," cr ;
   
   \  : output-c-func ( -- )
   \  \ used for word libraries
   \      ." Cell * I_" prim prim-c-name 2@ type ." (Cell *SP, Cell **FP)      /* " prim prim-name 2@ type
   \      ."  ( " prim prim-stack-string 2@ type ."  ) */" cr
   \      ." /* " prim prim-doc 2@ type ."  */" cr
   \      ." NAME(" quote prim prim-name 2@ type quote ." )" cr
   \      \ debugging
   \      ." {" cr
   \      print-declarations
   \      \ !! don't know what to do about that
   \      inst-stream  stack-used? IF ." Cell *ip=IP;" cr THEN
   \      data-stack   stack-used? IF ." Cell *sp=SP;" cr THEN
   \      fp-stack     stack-used? IF ." Cell *fp=*FP;" cr THEN
   \      return-stack stack-used? IF ." Cell *rp=*RP;" cr THEN
   \      spill-state
   \      fetches
   \      stack-pointer-updates
   \      fp-stack   stack-used? IF ." *FP=fp;" cr THEN
   \      ." {" cr
   \      ." #line " c-line @ . quote c-filename 2@ type quote cr
   \      prim prim-c-code 2@ type
   \      ." }" cr
   \      stores
   \      fill-state
   \      ." return (sp);" cr
   \      ." }" cr
   \      cr ;
   
   : output-label ( -- )  
       ." INST_ADDR(" prim prim-c-name 2@ type ." )," cr ;
   
   : output-alias ( -- ) 
       ( primitive-number @ . ." alias " ) ." Primitive " prim prim-name 2@ type cr ;
   
   defer output-c-prim-num ( -- )
   
   :noname ( -- )
       ." N_" prim prim-c-name 2@ type ." ," cr ;
   is output-c-prim-num
   
   : output-forth ( -- )  
       prim prim-forth-code @ 0=
       IF          \ output-alias
           \ this is bad for ec: an alias is compiled if tho word does not exist!
           \ JAW
       ELSE  ." : " prim prim-name 2@ type ."   ( "
           prim prim-stack-string 2@ type ." )" cr
           prim prim-forth-code 2@ type cr
       THEN ;
   
   : output-tag-file ( -- )
       name-filename 2@ last-name-filename 2@ compare if
           name-filename 2@ last-name-filename 2!
           #ff emit cr
           name-filename 2@ type
           ." ,0" cr
       endif ;
   
   : output-tag ( -- )
       output-tag-file
       prim prim-name 2@ 1+ type
       127 emit
       space prim prim-name 2@ type space
       1 emit
       name-line @ 0 .r
       ." ,0" cr ;
   
   : output-vi-tag ( -- )
       name-filename 2@ type #tab emit
       prim prim-name 2@ type #tab emit
       ." /^" prim prim-name 2@ type ."  *(/" cr ;
   
 [IFDEF] documentation  [IFDEF] documentation
 : register-doc ( -- )  : register-doc ( -- )
     get-current documentation set-current      prim prim-name 2@ documentation ['] create insert-wordlist
     forth-name 2@ nextname create      prim prim-name 2@ 2,
     forth-name 2@ 2,      prim prim-stack-string 2@ condition-stack-effect 2,
     stack-string 2@ condition-stack-effect 2,      prim prim-wordset 2@ 2,
     wordset 2@ 2,      prim prim-c-name 2@ condition-pronounciation 2,
     c-name 2@ condition-pronounciation 2,      prim prim-doc 2@ 2, ;
     doc 2@ 2,  [THEN]
     set-current ;  
   
   \ combining instructions
   
   \ The input should look like this:
   
   \ lit_+ = lit +
   
   \ The output should look like this:
   
   \  I_lit_+:
   \  {
   \  DEF_CA
   \  Cell _x_ip0;
   \  Cell _x_sp0;
   \  Cell _x_sp1;
   \  NEXT_P0;
   \  _x_ip0 = (Cell) IPTOS;
   \  _x_sp0 = (Cell) spTOS;
   \  INC_IP(1);
   \  /* sp += 0; */
   \  /* lit ( #w -- w ) */
   \  /*  */
   \  NAME("lit")
   \  {
   \  Cell w;
   \  w = (Cell) _x_ip0;
   \  #ifdef VM_DEBUG
   \  if (vm_debug) {
   \  fputs(" w=", vm_out); printarg_w (w);
   \  fputc('\n', vm_out);
   \  }
   \  #endif
   \  {
   \  #line 136 "./prim"
   \  }
   \  _x_sp1 = (Cell)w;
   \  }
   \  I_plus:      /* + ( n1 n2 -- n ) */
   \  /*  */
   \  NAME("+")
   \  {
   \  DEF_CA
   \  Cell n1;
   \  Cell n2;
   \  Cell n;
   \  NEXT_P0;
   \  n1 = (Cell) _x_sp0;
   \  n2 = (Cell) _x_sp1;
   \  #ifdef VM_DEBUG
   \  if (vm_debug) {
   \  fputs(" n1=", vm_out); printarg_n (n1);
   \  fputs(" n2=", vm_out); printarg_n (n2);
   \  fputc('\n', vm_out);
   \  }
   \  #endif
   \  {
   \  #line 516 "./prim"
   \  n = n1+n2;
   \  }
   \  _x_sp0 = (Cell)n;
   \  }
   \  NEXT_P1;
   \  spTOS = (Cell)_x_sp0;
   \  NEXT_P2;
   
   : init-combined ( -- )
       prim to combined
       0 num-combined !
       current-depth max-stacks cells erase
       include-skipped-insts @ current-depth 0 th !
       max-depth     max-stacks cells erase
       min-depth     max-stacks cells erase
       prim prim-effect-in  prim prim-effect-in-end  !
       prim prim-effect-out prim prim-effect-out-end ! ;
   
   : max! ( n addr -- )
       tuck @ max swap ! ;
   
   : min! ( n addr -- )
       tuck @ min swap ! ;
   
   : inst-stream-adjustment ( nstack -- n )
       \ number of stack items to add for each part
       0= include-skipped-insts @ and negate ;
   
   : add-depths { p -- }
       \ combine stack effect of p with *-depths
       max-stacks 0 ?do
           current-depth i th @
           p prim-stacks-in  i th @ + i inst-stream-adjustment +
           dup max-depth i th max!
           p prim-stacks-out i th @ -
           dup min-depth i th min!
           current-depth i th !
       loop ;
   
   : copy-maxdepths ( n -- )
       max-depth max-depths rot max-stacks * th max-stacks cells move ;
   
   : add-prim ( addr u -- )
       \ add primitive given by "addr u" to combined-prims
       primitives search-wordlist s" unknown primitive" ?print-error
       execute { p }
       p combined-prims num-combined @ th !
       num-combined @ copy-maxdepths
       1 num-combined +!
       p add-depths
       num-combined @ copy-maxdepths ;
   
   : compute-effects { q -- }
       \ compute the stack effects of q from the depths
       max-stacks 0 ?do
           max-depth i th @ dup
           q prim-stacks-in i th !
           current-depth i th @ -
           q prim-stacks-out i th !
       loop ;
   
   : make-effect-items { stack# items effect-endp -- }
       \ effect-endp points to a pointer to the end of the current item-array
       \ and has to be updated
       stacks stack# th @ { stack }
       items 0 +do
           effect-endp @ { item }
           i 0 <# #s stack stack-pointer 2@ holds [char] _ hold #> save-mem
           item item-name 2!
           stack item item-stack !
           stack stack-type @ item item-type !
           i item item-offset !
           item item-first on
           item% %size effect-endp +!
       loop ;
   
   : init-effects { q -- }
       \ initialize effects field for FETCHES and STORES
       max-stacks 0 ?do
           i q prim-stacks-in  i th @ q prim-effect-in-end  make-effect-items
           i q prim-stacks-out i th @ q prim-effect-out-end make-effect-items
       loop ;
   
   : compute-stack-max-back-depths ( stack -- )
       stack-number @ { stack# }
       current-depth stack# th @ dup
       dup stack# num-combined @ s-c-max-back-depth !
       -1 num-combined @ 1- -do ( max-depth current-depth )
           combined-prims i th @ { p }
           p prim-stacks-out stack# th @ +
           dup >r max r>
           over stack# i s-c-max-back-depth !
           p prim-stacks-in stack# th @ -
           stack# inst-stream-adjustment -
       1 -loop
       assert( dup stack# inst-stream-adjustment negate = )
       assert( over max-depth stack# th @ = )
       2drop ;
   
   : compute-max-back-depths ( -- )
       \ compute max-back-depths.
       \ assumes that current-depths is correct for the end of the combination
       ['] compute-stack-max-back-depths map-stacks ;
   
   : process-combined ( -- )
       combined combined-prims num-combined @ cells
       combinations ['] constant insert-wordlist
       combined-prims num-combined @ 1- th ( last-part )
       @ prim-c-code 2@ prim prim-c-code 2! \ used by output-super-end
       prim compute-effects
       prim init-effects
       compute-max-back-depths
       output-combined perform ;
   
   \ reprocessing (typically to generate versions for another cache states)
   \ !! use prim-context
   
   variable reprocessed-num 0 reprocessed-num !
   
   : new-name ( -- c-addr u )
       reprocessed-num @ 0
       1 reprocessed-num +!
       <# #s 'p hold '_ hold #> save-mem ;
   
   : reprocess-simple ( prim -- )
       to prim
       new-name prim prim-c-name 2!
       output @ execute ;
   
   : lookup-prim ( c-addr u -- prim )
       primitives search-wordlist 0= -13 and throw execute ;
   
   : state-prim1 { in-state out-state prim -- }
       in-state out-state state-default dup d= ?EXIT
       in-state state-enabled? out-state state-enabled? and 0= ?EXIT
       in-state  to state-in
       out-state to state-out
       prim reprocess-simple ;
   
   : state-prim ( in-state out-state "name" -- )
       parse-word lookup-prim state-prim1 ;
   
   \ reprocessing with default states
   
   \ This is a simple scheme and should be generalized
   \ assumes we only cache one stack and use simple states for that
   
   0 value cache-stack  \ stack that we cache
   2variable cache-states \ states of the cache, starting with the empty state
   
   : compute-default-state-out ( n-in -- n-out )
       \ for the current prim
       cache-stack stack-in @ - 0 max
       cache-stack stack-prim-stacks-sync @ if
           drop 0
       endif
       cache-stack stack-out @ + cache-states 2@ nip 1- min ;
   
   : gen-prim-states ( prim -- )
       to prim
       cache-states 2@ swap { states } ( nstates )
       cache-stack stack-in @ +do
           states i th @
           states i compute-default-state-out th @
           prim state-prim1
       loop ;
   
   : prim-states ( "name" -- )
       parse-word lookup-prim gen-prim-states ;
   
   : gen-branch-states ( prim -- )
       \ generate versions that produce state-default; useful for branches
       to prim
       cache-states 2@ swap { states } ( nstates )
       cache-stack stack-in @ +do
           states i th @ state-default prim state-prim1
       loop ;
   
   : branch-states ( out-state "name" -- )
       parse-word lookup-prim gen-branch-states ;
   
   \ producing state transitions
   
   : gen-transitions ( "name" -- )
       parse-word lookup-prim { prim }
       cache-states 2@ { states nstates }
       nstates 0 +do
           nstates 0 +do
               i j <> if
                   states i th @ states j th @ prim state-prim1
               endif
           loop
       loop ;
   
   \ C output
   
   : print-item { n stack -- }
       \ print nth stack item name
       stack stack-type @ type-c-name 2@ type space
       ." MAYBE_UNUSED _" stack stack-pointer 2@ type n 0 .r ;
   
   : print-declarations-combined ( -- )
       max-stacks 0 ?do
           max-depth i th @ min-depth i th @ - 0 +do
               i stacks j th @ print-item ." ;" cr
           loop
       loop ;
   
   : part-fetches ( -- )
       fetches ;
   
   : part-output-c-tail ( -- )
       print-debug-results
       stores ;
   
   : output-combined-tail ( -- )
       part-output-c-tail
       in-part @ >r in-part off
       combined ['] output-c-tail-no-stores prim-context
       r> in-part ! ;
   
   : part-stack-pointer-updates ( -- )
       next-stack-number @ 0 +do
           i part-num @ 1+ s-c-max-depth @ dup
           i num-combined @ s-c-max-depth @ =    \ final depth
           swap i part-num @ s-c-max-depth @ <> \ just reached now
           part-num @ 0= \ first part
           or and if
               stacks i th @ stack-pointer-update
           endif
       loop ;
   
   : output-part ( p -- )
       to prim
       ." /* " prim prim-name 2@ prim-type ."  ( " prim prim-stack-string 2@ type ." ) */" cr
       ." NAME(" quote prim prim-name 2@ type quote ." )" cr \ debugging
       ." {" cr
       print-declarations
       part-fetches
       print-debug-args
       combined ['] part-stack-pointer-updates prim-context
       1 part-num +!
       prim add-depths \ !! right place?
       prim prim-c-code 2@ ['] output-combined-tail type-c-code
       part-output-c-tail
       ." }" cr ;
   
   : output-parts ( -- )
       prim >r in-part on
       current-depth max-stacks cells erase
       0 part-num !
       ['] output-part map-combined
       in-part off
       r> to prim ;
   
   : output-c-combined ( -- )
       print-entry cr
       \ debugging messages just in parts
       ." {" cr
       ." DEF_CA" cr
       print-declarations-combined
       output-nextp0
       spill-state
       \ fetches \ now in parts
       \ print-debug-args
       \ stack-pointer-updates now in parts
       output-parts
       output-c-tail2-no-stores
       ." }" cr
       cr ;
   
   : output-forth-combined ( -- )
   ;
   
   
   \ peephole optimization rules
   
   \ data for a simple peephole optimizer that always tries to combine
   \ the currently compiled instruction with the last one.
   
   \ in order for this to work as intended, shorter combinations for each
   \ length must be present, and the longer combinations must follow
   \ shorter ones (this restriction may go away in the future).
     
   : output-peephole ( -- )
       combined-prims num-combined @ 1- cells combinations search-wordlist
       s" the prefix for this superinstruction must be defined earlier" ?print-error
       ." {"
       execute prim-num @ 5 .r ." ,"
       combined-prims num-combined @ 1- th @ prim-num @ 5 .r ." ,"
       combined prim-num @ 5 .r ." }, /* "
       combined prim-c-name 2@ type ."  */"
       cr ;
   
   
   \ cost and superinstruction data for a sophisticated combiner (e.g.,
   \ shortest path)
   
   \ This is intended as initializer for a structure like this
   
   \  struct cost {
   \    char loads;       /* number of stack loads */
   \    char stores;      /* number of stack stores */
   \    char updates;     /* number of stack pointer updates */
   \    char branch;      /* is it a branch (SET_IP) */
   \    char state_in;    /* state on entry */
   \    char state_out;   /* state on exit */
   \    short offset;     /* offset into super2 table */
   \    char length;      /* number of components */
   \  };
   
   \ How do you know which primitive or combined instruction this
   \ structure refers to?  By the order of cost structures, as in most
   \ other cases.
   
   : super2-length ( -- n )
       combined if
           num-combined @
       else
           1
       endif ;
   
   : compute-costs { p -- nloads nstores nupdates }
       \ compute the number of loads, stores, and stack pointer updates
       \ of a primitive or combined instruction; does not take TOS
       \ caching into account
       0 max-stacks 0 +do
           p prim-stacks-in i th @ +
       loop
       super2-length 1- - \ don't count instruction fetches of subsumed insts
       0 max-stacks 0 +do
           p prim-stacks-out i th @ +
       loop
       0 max-stacks 1 +do \ don't count ip updates, therefore "1 +do"
           p prim-stacks-in i th @ p prim-stacks-out i th @ <> -
       loop ;
   
   : output-num-part ( p -- )
       ." N_" prim-c-name-orig 2@ type ." ," ;
       \ prim-num @ 4 .r ." ," ;
   
   : output-name-comment ( -- )
       ."  /* " prim prim-name 2@ prim-type ."  */" ;
   
   variable offset-super2  0 offset-super2 ! \ offset into the super2 table
   
   : output-costs-prefix ( -- )
       ." {" prim compute-costs
       rot 2 .r ." ," swap 2 .r ." ," 2 .r ." , "
       prim prim-branch? negate . ." ,"
       state-in  state-number @ 2 .r ." ,"
       state-out state-number @ 2 .r ." ,"
       inst-stream stack-in @ 1 .r ." ,"
   ;
   
   : output-costs-gforth-simple ( -- )
       output-costs-prefix
       prim output-num-part
       1 2 .r ." },"
       output-name-comment
       cr ;
   
   : output-costs-gforth-combined ( -- )
       output-costs-prefix
       ." N_START_SUPER+" offset-super2 @ 5 .r ." ,"
       super2-length dup 2 .r ." }," offset-super2 +!
       output-name-comment
       cr ;
   
   \  : output-costs ( -- )
   \      \ description of superinstructions and simple instructions
   \      ." {" prim compute-costs
   \      rot 2 .r ." ," swap 2 .r ." ," 2 .r ." ,"
   \      offset-super2 @ 5 .r ." ,"
   \      super2-length dup 2 .r ." ," offset-super2 +!
   \      inst-stream stack-in @ 1 .r ." },"
   \      output-name-comment
   \      cr ;
   
   : output-super2-simple ( -- )
       prim prim-c-name 2@ prim prim-c-name-orig 2@ d= if
           prim output-num-part
           output-name-comment
           cr
       endif ;   
     
   : output-super2-combined ( -- )
       ['] output-num-part map-combined 
       output-name-comment
       cr ;   
   
   \ the parser
   
   eof-char max-member \ the whole character set + EOF
   
   : getinput ( -- n )
    rawinput @ endrawinput @ =
    if
      eof-char
    else
      cookedinput @ c@
    endif ;
   
   :noname ( n -- )
    dup bl > if
     emit space
    else
     .
    endif ;
   print-token !
   
   : testchar? ( set -- f )
    getinput member? ;
   ' testchar? test-vector !
   
   : checksynclines ( -- )
       \ when input points to a newline, check if the next line is a
       \ sync line.  If it is, perform the appropriate actions.
       rawinput @ begin >r
           s" #line " r@ over compare if
               rdrop 1 line +! EXIT
           endif
           0. r> 6 chars + 20 >number drop >r drop line ! r> ( c-addr )
           dup c@ bl = if
               char+ dup c@ [char] " <> 0= s" sync line syntax" ?print-error
               char+ dup 100 [char] " scan drop swap 2dup - save-mem filename 2!
               char+
           endif
           dup c@ nl-char <> 0= s" sync line syntax" ?print-error
           skipsynclines @ if
               char+ dup rawinput !
               rawinput @ c@ cookedinput @ c!
           endif
       again ;
   
   : ?nextchar ( f -- )
       s" syntax error, wrong char" ?print-error
       rawinput @ endrawinput @ <> if
           rawinput @ c@
           1 chars rawinput +!
           1 chars cookedinput +!
           nl-char = if
               checksynclines
               rawinput @ line-start !
           endif
           rawinput @ c@
           cookedinput @ c!
       endif ;
   
   : charclass ( set "name" -- )
    ['] ?nextchar terminal ;
   
   : .. ( c1 c2 -- set )
    ( creates a set that includes the characters c, c1<=c<=c2 )
    empty copy-set
    swap 1+ rot do
     i over add-member
    loop ;
   
   : ` ( -- terminal ) ( use: ` c )
    ( creates anonymous terminal for the character c )
    char singleton ['] ?nextchar make-terminal ;
   
   char a char z ..  char A char Z ..  union char _ singleton union  charclass letter
   char 0 char 9 ..                                        charclass digit
   bl singleton tab-char over add-member                   charclass white
   nl-char singleton eof-char over add-member complement   charclass nonl
   nl-char singleton eof-char over add-member
       char : over add-member complement                   charclass nocolonnl
   nl-char singleton eof-char over add-member
       char } over add-member complement                   charclass nobracenl
   bl 1+ maxchar .. char \ singleton complement intersection
                                                           charclass nowhitebq
   bl 1+ maxchar ..                                        charclass nowhite
   char " singleton eof-char over add-member complement    charclass noquote
   nl-char singleton                                       charclass nl
   eof-char singleton                                      charclass eof
   nl-char singleton eof-char over add-member              charclass nleof
   
   (( letter (( letter || digit )) **
   )) <- c-ident ( -- )
   
   (( ` . ` . ` .
   )) <- sync-stack ( -- )
   
   (( ` # ?? (( letter || digit || ` : )) ++ sync-stack ??
   || sync-stack
   )) <- stack-ident ( -- )
   
   (( nowhitebq nowhite ** ))
   <- forth-ident ( -- )
   
   Variable forth-flag
   Variable c-flag
   
   (( (( ` e || ` E )) {{ start }} nonl ** 
      {{ end evaluate }}
   )) <- eval-comment ( ... -- ... )
   
   (( (( ` f || ` F )) {{ start }} nonl ** 
      {{ end forth-flag @ IF type cr ELSE 2drop THEN }}
   )) <- forth-comment ( -- )
   
   (( (( ` c || ` C )) {{ start }} nonl ** 
      {{ end c-flag @ IF type cr ELSE 2drop THEN }}
   )) <- c-comment ( -- )
   
   (( ` - nonl ** {{ 
           forth-flag @ IF forth-fdiff ." [ELSE]" cr THEN
           c-flag @ IF
               function-diff
               ." #else /* " function-number @ 0 .r ."  */" cr THEN }}
   )) <- else-comment
   
   (( ` + {{ start }} nonl ** {{ end
           dup
           IF
               c-flag @ IF
                   function-diff
                   ." #ifdef HAS_" 2dup bounds ?DO  I c@ toupper emit  LOOP cr
               THEN
               forth-flag @ IF
                   forth-fdiff  ." has? " 2dup type ."  [IF]"  cr
               THEN
               2drop
           ELSE
               2drop
               c-flag @      IF
                   function-diff  ." #endif" cr THEN
               forth-flag @  IF  forth-fdiff  ." [THEN]"  cr THEN
           THEN }}
   )) <- if-comment
   
   (( (( ` g || ` G )) {{ start }} nonl **
      {{ end
         forth-flag @ IF  forth-fdiff  ." group " 2dup type cr  THEN
         c-flag @     IF  function-diff
             ." GROUP(" 2dup type ." , " function-number @ 0 .r ." )" cr  THEN
         2drop }}
   )) <- group-comment
   
   (( (( eval-comment || forth-comment || c-comment || else-comment || if-comment || group-comment )) ?? nonl ** )) <- comment-body
   
   (( ` \ comment-body nleof )) <- comment ( -- )
   
   (( {{ start }} stack-ident {{ end init-item1 }} white ** )) **
   <- stack-items ( addr1 -- addr2 )
   
   (( {{ prim prim-effect-in }}  stack-items {{ prim prim-effect-in-end ! }}
      ` - ` - white **
      {{ prim prim-effect-out }} stack-items {{ prim prim-effect-out-end ! }}
   )) <- stack-effect ( -- )
   
   (( {{ prim create-prim prim init-simple }}
      ` ( white ** {{ start }} stack-effect {{ end prim prim-stack-string 2! }} ` ) white **
      (( {{ start }} forth-ident {{ end prim prim-wordset 2! }} white **
         (( {{ start }}  c-ident {{ end prim-c-name-2! }} )) ??
      )) ??  nleof
      (( ` " ` "  {{ start }} (( noquote ++ ` " )) ++ {{ end 1- prim prim-doc 2! }} ` " white ** nleof )) ??
      {{ skipsynclines off line @ c-line ! filename 2@ c-filename 2! start }}
      (( (( ` { nonl ** nleof (( (( nobracenl {{ line @ drop }} nonl ** )) ?? nleof )) ** ` } white ** nleof white ** ))
      || (( nocolonnl nonl **  nleof white ** )) ** ))
      {{ end prim prim-c-code 2! skipsynclines on }}
      (( ` :  white ** nleof
         {{ start }} (( nonl ++  nleof white ** )) ++ {{ end prim prim-forth-code 2! }}
      )) ?? {{ process-simple }}
      nleof
   )) <- simple-primitive ( -- )
   
   (( {{ init-combined }}
      ` = white ** (( {{ start }} forth-ident {{ end add-prim }} white ** )) ++
      nleof {{ process-combined }}
   )) <- combined-primitive
   
   (( {{ make-prim to prim 0 to combined
         line @ name-line ! filename 2@ name-filename 2!
         function-number @ prim prim-num !
         start }} [ifdef] vmgen c-ident [else] forth-ident [then] {{ end
         2dup prim prim-name 2! prim-c-name-2! }}  white **
      (( ` / white ** {{ start }} c-ident {{ end prim-c-name-2! }} white ** )) ??
      (( simple-primitive || combined-primitive ))
      {{ 1 function-number +! }}
   )) <- primitive ( -- )
   
   (( (( comment || primitive || nl white ** )) ** eof ))
   parser primitives2something
   warnings @ [IF]
   .( parser generated ok ) cr
 [THEN]  [THEN]
   
 : process-file ( addr u xt -- )  
  >r r/o open-file abort" cannot open file"  
  warnings @ if  
  ." ------------ CUT HERE -------------" cr  endif  
  r> primfilter ;  
   
   \ run with gforth-0.5.0 (slurp-file is missing)
   [IFUNDEF] slurp-file
   : slurp-file ( c-addr1 u1 -- c-addr2 u2 )
       \ c-addr1 u1 is the filename, c-addr2 u2 is the file's contents
       r/o bin open-file throw >r
       r@ file-size throw abort" file too large"
       dup allocate throw swap
       2dup r@ read-file throw over <> abort" could not read whole file"
       r> close-file throw ;
   [THEN]
   
   : primfilter ( addr u -- )
       \ process the string at addr u
       over dup rawinput ! dup line-start ! cookedinput !
       + endrawinput !
       checksynclines
       primitives2something ;    
   
   : unixify ( c-addr u1 -- c-addr u2 )
       \ delete crs from the string
       bounds tuck tuck ?do ( c-addr1 )
           i c@ dup #cr <> if
               over c! char+
           else
               drop
           endif
       loop
       over - ;
   
   : process-file ( addr u xt-simple x-combined -- )
       output-combined ! output !
       save-mem 2dup filename 2!
       slurp-file unixify
       warnings @ if
           ." ------------ CUT HERE -------------" cr  endif
       primfilter ;
   
   \  : process      ( xt -- )
   \      bl word count rot
   \      process-file ;

Removed from v.1.16  
changed lines
  Added in v.1.169


FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>