--- gforth/float.fs 2000/09/23 15:46:56 1.34 +++ gforth/float.fs 2002/10/27 09:57:10 1.37 @@ -102,7 +102,7 @@ DOES> ( -- r ) : f. ( r -- ) \ float-ext f-dot \G Display (the floating-point number) @i{r} without exponent, \G followed by a space. - f\$ dup >r 0< + f\$ dup >r 0<= IF '0 emit ELSE scratch r@ min type r@ precision - zeros THEN '. emit r@ negate zeros @@ -180,6 +180,9 @@ IS interpreter-notfound \G Divide 1.0e0 by @i{r1}. 1.0e0 fswap f/ ; +get-current environment-wordlist set-current +1.7976931348623157e308 FConstant max-float +set-current \ We now have primitives for these, so we need not define them @@ -219,6 +222,19 @@ IS interpreter-notfound fnegate f~rel THEN ; +2e 53e f** fconstant fround-offset \ !! does f** work correctly? + +: fround ( r1 -- r2 ) \ float f-round +\G Round to nearest integral value. Break ties with round-to-even. + \ assumes IEEE DP FP in round-to-even mode. + \ for an explanation of this code read + \ <2002Oct26.113823@a0.complang.tuwien.ac.at> ff. + fdup f0> if + fround-offset f- fround-offset f+ + else fdup f0< if \ leave 0e and -0e as is + fround-offset f+ fround-offset f- + then then ; + : f.s ( -- ) \ gforth f-dot-s \G Display the number of items on the floating-point stack, \G followed by a list of the items; TOS is the right-most item.