[Ecls-list] wrong value for (float-sign -0.0)

Mark Hoemmen mark.hoemmen at gmail.com
Sun Sep 28 17:38:40 UTC 2008


Gabriel Dos Reis wrote:
> Signed zeros are important for the semantics numerics based on IEEE
> 754 semantics.   The way I disovered was precisely because I was
> debugging a failing OpenAxiom program based using ECL as its Lisp.
> So far, SBCL which is the other Lisp using native `double' is the only
> free Lisp which gets it right (my version of CLisp does not pretend to
> have :IEEE-FLOATING-POINT on its *features*).

Just for reference,

W. Kahan,
"Branch Cuts for Complex Elementary Functions, or Much Ado About
Nothing's Sign Bit"
    in The State of the Art in Numerical Analysis, (eds. Iserles and
Powell), Clarendon Press, Oxford, 1987

explains the rational for +/-0.0.

mfh






More information about the ecl-devel mailing list