21 Oct 2002

bug fix in e_pow.c from "David G Hough at validlab.com" <validlab@validlab.com>.

credit for bug report goes to andrew_johnson@uk.ibm.com

test example:

-----------------------------

static int fail=0;

trypow(x,y,z)

double x,y,z;

{

extern double pow();

double p ;

p = pow(x,y);

if (p != z) {

        printf(" pow failure x %e y %e computed %e correct %e \n",x,y,p,z);

        fail++;

        return;

}

if (1/p != 1/z) {

        printf(" pow failure x %e y %e computed %e correct %e \n",x,y,p,z);

        fail++;

        return;

}

}

main()

{

trypow( 1.0000000001, 1.0000000001E10, 2.7182820535066154);

trypow(-1.0000000001, 1.0000000001E10, -2.7182820535066154);

trypow(-0.001, 1.0000000001E10, -0.0);

trypow(-1000.0, 1.0000000001E10, -1.0/0.0);

if (fail == 0) {

        printf(" fdlibm e_pow.c seems to be current \n");

        exit(0);

}

else {

        printf(" fdlibm e_pow.c seems to be out of date \n");

        exit(1);

}

}

-----------------------------

What's new in FDLIBM 5.3?

CONFIGURE

        To build FDLIBM, edit the supplied Makefile or create

        a local Makefile by running "sh configure"

        using the supplied configure script contributed by Nelson Beebe

BUGS FIXED

    1. e_pow.c incorrect results when

        x is very close to -1.0 and y is very large, e.g.

        pow(-1.0000000000000002e+00,4.5035996273704970e+15) = 0

        pow(-9.9999999999999978e-01,4.5035996273704970e+15) = 0

        Correct results are close to -e and -1/e.

    2. k_tan.c error was > 1 ulp target for FDLIBM

        5.2: Worst error at least 1.45 ulp at

        tan(1.7765241907548024E+269) = 1.7733884462610958E+16

        5.3: Worst error 0.96 ulp

NOT FIXED YET

    3. Compiler failure on non-standard code

        Statements like

                    *(1+(int*)&t1) = 0;

        are not standard C and cause some optimizing compilers (e.g. GCC)

        to generate bad code under optimization.    These cases

        are to be addressed in the next release.