Diff for /gforth/float.fs between versions 1.29 and 1.33

version 1.29, 1999/12/03 18:24:22 version 1.33, 2000/09/23 12:27:46
Line 51 Line 51
     \G enough space to align it.      \G enough space to align it.
     here dup dfaligned swap ?DO  bl c,  LOOP ;      here dup dfaligned swap ?DO  bl c,  LOOP ;
   
 1 sfloats constant sfloat+ ( sf-addr1 -- sf-addr2 ) \ float-ext s-float-plus  1 sfloats (Field) sfloat+ , ( sf-addr1 -- sf-addr2 ) \ float-ext s-float-plus
 \G Increment @i{sf-addr1} by the number of address units corresponding to the size of  \G @code{1 sfloats +}.
 \G a single-precision IEEE floating-point number, to give @i{sf-addr2}.""  
 dofield: lastxt code-address! \ change the constant into a field  1 dfloats (Field) dfloat+ , ( df-addr1 -- df-addr2 ) \ float-ext d-float-plus
   \G @code{1 dfloats +}.
 1 dfloats constant dfloat+ ( df-addr1 -- df-addr2 ) \ float-ext d-float-plus  
 \G Increment @i{df-addr1} by the number of address units corresponding to the size of  
 \G a double-precision IEEE floating-point number, to give @i{df-addr2}.""  
 dofield: lastxt code-address! \ change the constant into a field  
   
 : f, ( f -- ) \ gforth  : f, ( f -- ) \ gforth
     \G Reserve data space for one floating-point number and store      \G Reserve data space for one floating-point number and store
Line 104  DOES> ( -- r ) Line 100  DOES> ( -- r )
   IF  '- emit  THEN ;    IF  '- emit  THEN ;
   
 : f.  ( r -- ) \ float-ext f-dot  : f.  ( r -- ) \ float-ext f-dot
 \G Display (the floating-point number) @i{r} using fixed-point notation,  \G Display (the floating-point number) @i{r} without exponent,
 \G followed by a space.  \G followed by a space.
   f$ dup >r 0<    f$ dup >r 0<
   IF    '0 emit    IF    '0 emit
Line 114  DOES> ( -- r ) Line 110  DOES> ( -- r )
 \ I'm afraid this does not really implement ansi semantics wrt precision.  \ I'm afraid this does not really implement ansi semantics wrt precision.
 \ Shouldn't precision indicate the number of places shown after the point?  \ Shouldn't precision indicate the number of places shown after the point?
   
   \ Why do you think so? ANS Forth appears ambiguous on this point. -anton.
   
 : fe. ( r -- ) \ float-ext f-e-dot  : fe. ( r -- ) \ float-ext f-e-dot
 \G Display @i{r} using engineering notation, followed by a space.  \G Display @i{r} using engineering notation (with exponent dividable
   \G by 3), followed by a space.
   f$ 1- s>d 3 fm/mod 3 * >r 1+ >r    f$ 1- s>d 3 fm/mod 3 * >r 1+ >r
   scratch r@ min type '. emit  scratch r> /string type    scratch r@ tuck min tuck - >r type r> zeros
     '. emit scratch r> /string type
   'E emit r> . ;    'E emit r> . ;
   
 : fs. ( r -- ) \ float-ext f-s-dot  : fs. ( r -- ) \ float-ext f-s-dot
 \G Display @i{r} using scientific notation, followed by a space.  \G Display @i{r} using scientific notation (with exponent), followed
   \G by a space.
   f$ 1-    f$ 1-
   scratch over c@ emit '. emit 1 /string type    scratch over c@ emit '. emit 1 /string type
   'E emit . ;    'E emit . ;
Line 203  IS interpreter-notfound Line 204  IS interpreter-notfound
         f- fabs frot frot f* f< ;          f- fabs frot frot f* f< ;
   
 : f~ ( r1 r2 r3 -- flag ) \ float-ext f-proximate  : f~ ( r1 r2 r3 -- flag ) \ float-ext f-proximate
     \G ANS Forth medley: r3>0: @code{f~abs}; r3=0: r1=r2; r3<0: @code{fnegate f~abs}.      \G ANS Forth medley for comparing r1 and r2 for equality: r3>0:
       \G @code{f~abs}; r3=0: bitwise comparison; r3<0: @code{fnegate f~rel}.
     fdup f0=      fdup f0=
     IF      IF \ bitwise comparison
         fdrop f=  \ !! this does not work, because 0=-0 with f= on Linux-Intel          fp@ float+ 1 floats over float+ -text 0=
                   \ the standard says they should compare unequal          fdrop fdrop fdrop
                   \ the comparison should be done with COMPARE  
         EXIT          EXIT
     THEN      THEN
     fdup f0>      fdup f0>

Removed from v.1.29  
changed lines
  Added in v.1.33


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