Re: [Chicken-users] IEEE float arithmetic

[John Cowan]

felix winkelmann scripsit:

> So you have 2.320 running ok, yes? Do you get this on installation or 
> build
> time? If the former, touch *.scm and run make again (without 
> BOOTSTRAP_PATH).
> Really weird, this. Or you can try the chicken.egg
> ("chicken-setup -f chicken" and extract/install by hand).

Okay.  I reinstalled 2.315 from the tarball, fetched the darcs head,
and did the autogen/configure/make/install dance.  That gave me a working
2.321 build.

However, the problems with +inf, -inf, +inf.0, -inf.0, and +nan persist:
all are undefined variables.  Only +nan works, and it still prints out
as NaN.0.

So I wrote a program that did a scanf("%lf", &d) and a printf, and all
the above forms were accepted on Linux but not on Cygwin.  I think this
must be due to Cygwin's use of newlib rather than GNU libc.  However, the
printf output is consistently "inf", "-inf", and "nan" on both systems.

Can you install a workaround for Cygwin that processes infs and nans
correctly on input, and also correct the output on all systems so we don't
get the non-inputtable forms Infinity.0, -Infinity.0, and +NaN.0 printed?
(I presume the problems are in string->number and number->string.)

-- 
What has four pairs of pants, lives             John Cowan
in Philadelphia, and it never rains             http://www.ccil.org/~cowan
but it pours?                                   address@hidden
        --Rufus T. Firefly