From: Kbwms AT aol DOT com Message-ID: <4d.3426682d.2c84f4ee@aol.com> Date: Mon, 1 Sep 2003 15:15:58 EDT Subject: Bug in log10l in libstdcxx.a To: djgpp-workers AT delorie DOT com MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="part1_4d.3426682d.2c84f4ee_boundary" X-Mailer: 8.0 for Windows sub 6015 Reply-To: djgpp-workers AT delorie DOT com --part1_4d.3426682d.2c84f4ee_boundary Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit I discovered this during checkout of my long double stuff via the test-vector suite. There is a library called libstdcxx.a in ../lib that gets linked with my C++ code. It is producing erroneous results as shown below: Tests using log10l in libstdcxx.a ------------TEST 1------------ log10l(LDBL_MIN) = log10l(+3.3621031431120935063e-4932) = -Inf log10l(LDBL_MAX) = log10l(+1.189731495357231765e+4932) = +Inf ------------TEST 2------------ log10l(LDBL_MIN/2) = log10l(+1.6810515715560467531e-4932) = -Inf log10l(LDBL_MAX/2) = log10l(+5.948657476786158825e+4931) = +Inf ------------------------------------------------------------------------------ - The corrects results are: Tests using log10l in libongdbl.a ------------TEST 1------------ log10l(LDBL_MIN) = log10l(+3.3621031431120935063e-4932) = -4931.4733889673399401 log10l(LDBL_MAX) = log10l(+1.189731495357231765e+4932) = +4932.0754489586679026 ------------TEST 2------------ log10l(LDBL_MIN/2) = log10l(+1.6810515715560467531e-4932) = -4931.7744189630039213 log10l(LDBL_MAX/2) = log10l(+5.948657476786158825e+4931) = +4931.7744189630039213 KB Williams --part1_4d.3426682d.2c84f4ee_boundary Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable I discovered this during checkout of my long double stuf= f via the test-vector suite.  There is a library called libstdcxx.a in=20= ../lib that gets linked with my C++ code.  It is producing erroneous re= sults as shown below:

Tests using log10l in libstdcxx.a
------------TEST   1------------
log10l(LDBL_MIN) =3D log10l(+3.3621031431120935063e-4932) =3D -Inf
log10l(LDBL_MAX) =3D log10l(+1.189731495357231765e+4932) =3D +Inf
------------TEST   2------------
log10l(LDBL_MIN/2) =3D log10l(+1.6810515715560467531e-4932) =3D -Inf
log10l(LDBL_MAX/2) =3D log10l(+5.948657476786158825e+4931) =3D +Inf
----------------------------------------------------------------------------= ---

The corrects results are:


Tests using log10l in libongdbl.a
------------TEST   1------------
log10l(LDBL_MIN) =3D
log10l(+3.3621031431120935063e-4932) =3D -4931.4733889673399401

log10l(LDBL_MAX) =3D
log10l(+1.189731495357231765e+4932) =3D +4932.0754489586679026
------------TEST   2------------
log10l(LDBL_MIN/2) =3D
log10l(+1.6810515715560467531e-4932) =3D -4931.7744189630039213

log10l(LDBL_MAX/2) =3D
         log10l(+5.94865747678615882= 5e+4931) =3D +4931.7744189630039213


KB Williams --part1_4d.3426682d.2c84f4ee_boundary--