That was my other thought. I guess I was expecting the number to be rounded or truncated, but I will review the IEEE standard.
-Kevin On Thu, Dec 9, 2021 at 2:44 AM Alexander Burger <a...@software-lab.de> wrote: > Hi Kevin, > > > I'm getting strange results with the following: > > ... > > : (scl 16) > > -> 16 > > : (setq Ptr (%@ "malloc" 'P 4)) > > -> 9674752 > > : (struct Ptr NIL (list -1.0 1280.0)) > > -> NIL > > : (format (car (struct Ptr '(-1.0 . 1))) *Scl) > > -> "-103.5058256155770880" > > ... > > The value seems completely off; the examples for 'struct work fine. > > It is a floating point range problem. > > > The IEEE 'float' type cannot support such high precision (16 digits). > > If you do > > (setq Ptr (%@ "malloc" 'P 4)) > (scl 7) > (format (struct Ptr -1.0 (-1.0 123.0)) *Scl) > -> "123.0000000" > > it works fine, but for higher scales it more and more disintegrates, as in > your > example above. > > > Using IEEE 'double' helps for a while > > (setq Ptr (%@ "malloc" 'P 8)) > (scl 15) > (format (struct Ptr 1.0 (1.0 123.0)) *Scl) > -> "123.000000000000000" > > but no longer for 16 > > (scl 16) > (format (struct Ptr 1.0 (1.0 123.0)) *Scl) > -> "7.7078495393153024" > > ☺/ A!ex > > -- > UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe >
