summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--cbrt.c9
-rw-r--r--const_euler.c13
-rw-r--r--const_log2.c3
-rw-r--r--cos.c8
-rw-r--r--cosh.c19
-rw-r--r--div_ui.c7
-rw-r--r--eq.c21
-rw-r--r--exp.c7
-rw-r--r--exp_2.c23
-rw-r--r--get_ld.c3
-rw-r--r--hypot.c26
-rw-r--r--init2.c3
-rw-r--r--log.c27
-rw-r--r--log10.c17
-rw-r--r--log2.c27
-rw-r--r--reldiff.c14
-rw-r--r--save_expo.c11
-rw-r--r--set_f.c11
-rw-r--r--set_prc_raw.c14
-rw-r--r--set_prec.c5
-rw-r--r--set_str.c17
-rw-r--r--ui_div.c5
-rw-r--r--zeta.c3
23 files changed, 125 insertions, 168 deletions
diff --git a/cbrt.c b/cbrt.c
index eae5d71dd..bd7d15177 100644
--- a/cbrt.c
+++ b/cbrt.c
@@ -1,6 +1,6 @@
/* mpfr_cbrt -- cube root function.
-Copyright 2002, 2003 Free Software Foundation.
+Copyright 2002, 2003, 2004 Free Software Foundation.
Contributed by the Spaces project, INRIA Lorraine.
This file is part of the MPFR Library.
@@ -20,7 +20,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include <stdlib.h>
#include "gmp.h"
#include "gmp-impl.h"
@@ -46,7 +45,7 @@ MA 02111-1307, USA. */
*/
int
-mpfr_cbrt (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
+mpfr_cbrt (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
{
mpz_t m;
@@ -78,7 +77,7 @@ mpfr_cbrt (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
else
MPFR_ASSERTN(0);
}
- /* Useless due to mpz_init
+ /* Useless due to mpz_init
MPFR_CLEAR_FLAGS(y);*/
mpz_init (m);
@@ -95,7 +94,7 @@ mpfr_cbrt (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
n = MPFR_PREC(y);
if (rnd_mode == GMP_RNDN)
n ++;
-
+
/* we want 3*n-2 <= size_m + 3*sh + r <= 3*n
i.e. 3*sh + size_m + r <= 3*n */
sh = (3 * n - size_m - r) / 3;
diff --git a/const_euler.c b/const_euler.c
index ee1fdb8c6..c62ec78ec 100644
--- a/const_euler.c
+++ b/const_euler.c
@@ -1,6 +1,6 @@
/* mpfr_const_euler -- Euler's constant
-Copyright 2001, 2002 Free Software Foundation.
+Copyright 2001, 2002, 2004 Free Software Foundation.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include <stdlib.h>
#include "gmp.h"
#include "gmp-impl.h"
@@ -115,7 +114,7 @@ mpfr_const_euler_S (mpfr_t x, unsigned long n)
mpz_clear (t);
}
-/* computes R(n) = exp(-n)/n * sum(k!/(-n)^k, k=0..n-2)
+/* computes R(n) = exp(-n)/n * sum(k!/(-n)^k, k=0..n-2)
with error at most 4*ulp(x). Assumes n>=2.
Since x <= exp(-n)/n <= 1/8, then 4*ulp(x) <= ulp(1).
*/
@@ -137,7 +136,7 @@ mpfr_const_euler_R (mpfr_t x, unsigned long n)
for (k = 1; k <= n; k++)
{
- mpz_mul_ui (a, a, k);
+ mpz_mul_ui (a, a, k);
mpz_div_ui (a, a, n);
/* the error e(k) on a is e(k) <= 1 + k/n*e(k-1) with e(0)=0,
i.e. e(k) <= k */
@@ -148,11 +147,7 @@ mpfr_const_euler_R (mpfr_t x, unsigned long n)
}
/* the error on s is at most 1+2+...+n = n*(n+1)/2 */
mpz_div_ui (s, s, n); /* err <= 1 + (n+1)/2 */
- if (MPFR_PREC(x) < mpz_sizeinbase(s, 2))
- {
- fprintf (stderr, "prec(x) is too small in mpfr_const_euler_R\n");
- exit (1);
- }
+ MPFR_ASSERTN (MPFR_PREC(x) >= mpz_sizeinbase(s, 2));
mpfr_set_z (x, s, GMP_RNDD); /* exact */
mpfr_div_2ui (x, x, m, GMP_RNDD);
/* now x = 1/n * sum(k!/(-n)^k, k=0..n-2) <= 1/n */
diff --git a/const_log2.c b/const_log2.c
index 642173009..df00e9c98 100644
--- a/const_log2.c
+++ b/const_log2.c
@@ -1,6 +1,6 @@
/* mpfr_const_log2 -- compute natural logarithm of 2
-Copyright 1999, 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 1999, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "longlong.h"
diff --git a/cos.c b/cos.c
index 1b4015150..328061615 100644
--- a/cos.c
+++ b/cos.c
@@ -1,6 +1,6 @@
/* mpfr_cos -- cosine of a floating-point number
-Copyright 2001, 2002, 2003 Free Software Foundation.
+Copyright 2001, 2002, 2003, 2004 Free Software Foundation.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -27,8 +26,8 @@ MA 02111-1307, USA. */
static int mpfr_cos2_aux _MPFR_PROTO ((mpfr_ptr, mpfr_srcptr));
-int
-mpfr_cos (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
+int
+mpfr_cos (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
{
int K0, K, precy, m, k, l, inexact;
mpfr_t r, s;
@@ -138,4 +137,3 @@ mpfr_cos2_aux (mpfr_ptr s, mpfr_srcptr r)
return l;
}
-
diff --git a/cosh.c b/cosh.c
index 7964535ab..d6ff83bc3 100644
--- a/cosh.c
+++ b/cosh.c
@@ -1,6 +1,6 @@
/* mpfr_cosh -- hyperbolic cosine
-Copyright 2001, 2002 Free Software Foundation.
+Copyright 2001, 2002, 2004 Free Software Foundation.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -31,9 +30,9 @@ MA 02111-1307, USA. */
*/
int
-mpfr_cosh (mpfr_ptr y, mpfr_srcptr xt , mp_rnd_t rnd_mode)
+mpfr_cosh (mpfr_ptr y, mpfr_srcptr xt , mp_rnd_t rnd_mode)
{
-
+
/****** Declaration ******/
mpfr_t x;
mp_prec_t Nxt = MPFR_PREC(xt);
@@ -41,13 +40,13 @@ mpfr_cosh (mpfr_ptr y, mpfr_srcptr xt , mp_rnd_t rnd_mode)
if (MPFR_UNLIKELY(MPFR_IS_SINGULAR(xt)))
{
- if (MPFR_IS_NAN(xt))
+ if (MPFR_IS_NAN(xt))
{
- MPFR_SET_NAN(y);
+ MPFR_SET_NAN(y);
MPFR_RET_NAN;
}
else if (MPFR_IS_INF(xt))
- {
+ {
MPFR_SET_INF(y);
MPFR_SET_POS(y);
MPFR_RET(0);
@@ -65,7 +64,7 @@ mpfr_cosh (mpfr_ptr y, mpfr_srcptr xt , mp_rnd_t rnd_mode)
/* General case */
{
/* Declaration of the intermediary variable */
- mpfr_t t, te,ti;
+ mpfr_t t, te,ti;
/* Declaration of the size variable */
mp_prec_t Nx = Nxt; /* Precision of input variable */
@@ -73,7 +72,7 @@ mpfr_cosh (mpfr_ptr y, mpfr_srcptr xt , mp_rnd_t rnd_mode)
mp_prec_t Nt; /* Precision of the intermediary variable */
long int err; /* Precision of error */
-
+
/* compute the precision of intermediary variable */
Nt = MAX(Nx, Ny);
/* the optimal number of bits : see algorithms.ps */
@@ -110,7 +109,7 @@ mpfr_cosh (mpfr_ptr y, mpfr_srcptr xt , mp_rnd_t rnd_mode)
}
while ((err < 0) || !mpfr_can_round (t, err, GMP_RNDN, GMP_RNDZ,
Ny + (rnd_mode == GMP_RNDN)));
-
+
inexact = mpfr_set (y, t, rnd_mode);
mpfr_clear (t);
diff --git a/div_ui.c b/div_ui.c
index 729d15306..2c755311c 100644
--- a/div_ui.c
+++ b/div_ui.c
@@ -1,6 +1,6 @@
/* mpfr_div_ui -- divide a floating-point number by a machine integer
-Copyright 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "longlong.h"
@@ -43,8 +42,8 @@ mpfr_div_ui (mpfr_ptr y, mpfr_srcptr x, unsigned long int u, mp_rnd_t rnd_mode)
MPFR_SET_NAN(y);
MPFR_RET_NAN;
}
- else if (MPFR_IS_INF(x))
- {
+ else if (MPFR_IS_INF(x))
+ {
MPFR_SET_INF(y);
MPFR_SET_SAME_SIGN(y, x);
MPFR_RET(0);
diff --git a/eq.c b/eq.c
index 0771aa585..54d18e87a 100644
--- a/eq.c
+++ b/eq.c
@@ -1,6 +1,6 @@
/* mpfr_eq -- Compare two floats up to a specified bit #.
-Copyright 1999, 2001, 2003 Free Software Foundation, Inc.
+Copyright 1999, 2001, 2003, 2004 Free Software Foundation, Inc.
(Copied from GNU MP, file mpf_eq.)
This file is part of the MPFR Library.
@@ -20,7 +20,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -69,9 +68,9 @@ mpfr_eq (mpfr_srcptr u, mpfr_srcptr v, unsigned long int n_bits)
{
if ((unsigned long) vsize * BITS_PER_MP_LIMB < n_bits)
{
- k = usize - vsize - 1;
- while (k >= 0 && !up[k]) --k;
- if (k >= 0)
+ k = usize - vsize - 1;
+ while (k >= 0 && !up[k]) --k;
+ if (k >= 0)
return 0; /* surely too different */
}
size = vsize;
@@ -80,9 +79,9 @@ mpfr_eq (mpfr_srcptr u, mpfr_srcptr v, unsigned long int n_bits)
{
if ((unsigned long) usize * BITS_PER_MP_LIMB < n_bits)
{
- k = vsize - usize - 1;
- while (k >= 0 && !vp[k]) --k;
- if (k >= 0)
+ k = vsize - usize - 1;
+ while (k >= 0 && !vp[k]) --k;
+ if (k >= 0)
return 0; /* surely too different */
}
size = usize;
@@ -113,8 +112,8 @@ mpfr_eq (mpfr_srcptr u, mpfr_srcptr v, unsigned long int n_bits)
return mpfr_cmp (u, v) == 0;
if (n_bits & (BITS_PER_MP_LIMB - 1))
- return (up[i] >> (BITS_PER_MP_LIMB - (n_bits & (BITS_PER_MP_LIMB - 1))) ==
- vp[i] >> (BITS_PER_MP_LIMB - (n_bits & (BITS_PER_MP_LIMB - 1))));
+ return (up[i] >> (BITS_PER_MP_LIMB - (n_bits & (BITS_PER_MP_LIMB - 1))) ==
+ vp[i] >> (BITS_PER_MP_LIMB - (n_bits & (BITS_PER_MP_LIMB - 1))));
else
- return (up[i] == vp[i]);
+ return (up[i] == vp[i]);
}
diff --git a/exp.c b/exp.c
index 3dc710b30..3c5eb9928 100644
--- a/exp.c
+++ b/exp.c
@@ -1,6 +1,6 @@
/* mpfr_exp -- exponential of a floating-point number
-Copyright 1999, 2000, 2001, 2002, 2003 Free Software Foundation.
+Copyright 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation.
Contributed by the Spaces project.
This file is part of the MPFR Library.
@@ -20,7 +20,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -32,8 +31,8 @@ MA 02111-1307, USA. */
where x = n*log(2)+(2^K)*r
number of operations = O(K+prec(r)/K)
*/
-int
-mpfr_exp (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
+int
+mpfr_exp (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
{
int expx, precy;
double d;
diff --git a/exp_2.c b/exp_2.c
index 67f56c528..a82ced00d 100644
--- a/exp_2.c
+++ b/exp_2.c
@@ -1,7 +1,7 @@
-/* mpfr_exp_2 -- exponential of a floating-point number
+/* mpfr_exp_2 -- exponential of a floating-point number
using Brent's algorithms in O(n^(1/2)*M(n)) and O(n^(1/3)*M(n))
-Copyright 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -20,7 +20,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "longlong.h" /* for count_leading_zeros */
@@ -105,8 +104,8 @@ mpz_normalize2 (mpz_t rop, mpz_t z, int expz, int target)
together with Brent-Kung O(t^(1/2)) algorithm for the evaluation of
power series. The resulting complexity is O(n^(1/3)*M(n)).
*/
-int
-mpfr_exp_2 (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
+int
+mpfr_exp_2 (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
{
int n, K, precy, q, k, l, err, exps, inexact, error_r;
mpfr_t r, s, t; mpz_t ss;
@@ -162,7 +161,7 @@ mpfr_exp_2 (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
printf ("r="); mpfr_print_binary (r); putchar ('\n');
#endif
mpfr_sub (r, x, r, GMP_RNDU);
- /* possible cancellation here: the error on r is at most
+ /* possible cancellation here: the error on r is at most
3*2^(EXP(old_r)-EXP(new_r)) */
if (MPFR_SIGN(r) < 0)
{ /* initial approximation n was too large */
@@ -237,7 +236,7 @@ mpfr_exp_2 (mpfr_ptr y, mpfr_srcptr x, mp_rnd_t rnd_mode)
using naive method with O(l) multiplications.
Return the number of iterations l.
The absolute error on s is less than 3*l*(l+1)*2^(-q).
- Version using fixed-point arithmetic with mpz instead
+ Version using fixed-point arithmetic with mpz instead
of mpfr for internal computations.
s must have at least qn+1 limbs (qn should be enough, but currently fails
since mpz_mul_2exp(s, s, q-1) reallocates qn+1 limbs)
@@ -251,7 +250,7 @@ mpfr_exp2_aux (mpz_t s, mpfr_srcptr r, int q, int *exps)
TMP_MARK(marker);
qn = 1 + (q-1)/BITS_PER_MP_LIMB;
- MY_INIT_MPZ(t, 2*qn+1); /* 2*qn+1 is needed since mpz_div_2exp may
+ MY_INIT_MPZ(t, 2*qn+1); /* 2*qn+1 is needed since mpz_div_2exp may
need an extra limb */
MY_INIT_MPZ(rr, qn+1);
mpz_set_ui(t, 1); expt=0;
@@ -285,7 +284,7 @@ mpfr_exp2_aux (mpz_t s, mpfr_srcptr r, int q, int *exps)
/* s <- 1 + r/1! + r^2/2! + ... + r^l/l! while MPFR_EXP(r^l/l!)+MPFR_EXPR(r)>-q
using Brent/Kung method with O(sqrt(l)) multiplications.
Return l.
- Uses m multiplications of full size and 2l/m of decreasing size,
+ Uses m multiplications of full size and 2l/m of decreasing size,
i.e. a total equivalent to about m+l/m full multiplications,
i.e. 2*sqrt(l) for m=sqrt(l).
Version using mpz. ss must have at least (sizer+1) limbs.
@@ -343,9 +342,9 @@ mpfr_exp2_aux2 (mpz_t s, mpfr_srcptr r, int q, int *exps)
expR[i] = mpz_normalize2(R[i], R[i], expR[i], 1-ql);
}
/* the absolute error on R[i]*rr is still 2*i-1 ulps */
- expt = mpz_normalize2(t, R[m-1], expR[m-1], 1-ql);
+ expt = mpz_normalize2(t, R[m-1], expR[m-1], 1-ql);
/* err(t) <= 2*m-1 ulps */
- /* computes t = 1 + r/(l+1) + ... + r^(m-1)*l!/(l+m-1)!
+ /* computes t = 1 + r/(l+1) + ... + r^(m-1)*l!/(l+m-1)!
using Horner's scheme */
for (i=m-2;i>=0;i--)
{
@@ -368,7 +367,7 @@ mpfr_exp2_aux2 (mpz_t s, mpfr_srcptr r, int q, int *exps)
#endif
mpz_add(s, s, t); /* no error here */
- /* updates rr, the multiplication of the factors l+i could be done
+ /* updates rr, the multiplication of the factors l+i could be done
using binary splitting too, but it is not sure it would save much */
mpz_mul(t, rr, R[m]); /* err(t) <= err(rr) + 2m-1 */
expr += expR[m];
diff --git a/get_ld.c b/get_ld.c
index 0cca171ff..0be458d9a 100644
--- a/get_ld.c
+++ b/get_ld.c
@@ -1,7 +1,7 @@
/* mpfr_get_ld -- convert a multiple precision floating-point number
to a machine long double
-Copyright 2002, 2003 Free Software Foundation, Inc.
+Copyright 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -20,7 +20,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include <float.h>
#include "gmp.h"
diff --git a/hypot.c b/hypot.c
index 93a3ef937..bd5b43704 100644
--- a/hypot.c
+++ b/hypot.c
@@ -1,6 +1,6 @@
/* mpfr_hypot -- Euclidean distance
-Copyright 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,8 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
-#include <stdlib.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -32,7 +30,7 @@ MA 02111-1307, USA. */
*/
int
-mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
+mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
{
int inexact;
/* Flag exact computation */
@@ -58,7 +56,7 @@ mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
MPFR_RET(0);
}
else if (MPFR_IS_ZERO(x))
- return mpfr_abs (z, y, rnd_mode);
+ return mpfr_abs (z, y, rnd_mode);
else if (MPFR_IS_ZERO(y))
return mpfr_abs (z, x, rnd_mode);
else
@@ -84,7 +82,7 @@ mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
/* we have x < 2^Ex thus x^2 < 2^(2*Ex),
and ulp(x) = 2^(Ex-Nx) thus ulp(x^2) >= 2^(2*Ex-2*Nx).
- y does not overlap with the result when
+ y does not overlap with the result when
x^2+y^2 < (|x| + 1/2*ulp(x,Nz))^2 = x^2 + |x|*ulp(x,Nz) + 1/4*ulp(x,Nz)^2,
i.e. a sufficient condition is y^2 < |x|*ulp(x,Nz),
or 2^(2*Ey) <= 2^(2*Ex-1-Nz), i.e. 2*diff_exp > Nz
@@ -109,7 +107,7 @@ mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
Nx = MPFR_PREC(x); /* Precision of input variable */
Ny = MPFR_PREC(y); /* Precision of input variable */
-
+
/* compute the working precision -- see algorithms.ps */
Nt = MAX(MAX(MAX(Nx, Ny), Nz), 8);
Nt = Nt - 8 + __gmpfr_ceil_log2 (Nt);
@@ -125,13 +123,13 @@ mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
do
{
- Nt += 10;
+ Nt += 10;
not_exact = 0;
/* reactualization of the precision */
- mpfr_set_prec (t, Nt);
- mpfr_set_prec (te, Nt);
- mpfr_set_prec (ti, Nt);
+ mpfr_set_prec (t, Nt);
+ mpfr_set_prec (te, Nt);
+ mpfr_set_prec (ti, Nt);
/* computations of hypot */
mpfr_div_2ui (te, x, sh, GMP_RNDZ); /* exact since Nt >= Nx */
@@ -140,14 +138,14 @@ mpfr_hypot (mpfr_ptr z, mpfr_srcptr x , mpfr_srcptr y , mp_rnd_t rnd_mode)
mpfr_div_2ui (ti, y, sh, GMP_RNDZ); /* exact since Nt >= Ny */
if (mpfr_mul (ti, ti, ti, GMP_RNDZ)) /* y^2 */
- not_exact = 1;
-
+ not_exact = 1;
+
if (mpfr_add (t, te, ti, GMP_RNDZ)) /* x^2+y^2 */
not_exact = 1;
if (mpfr_sqrt (t, t, GMP_RNDZ)) /* sqrt(x^2+y^2)*/
not_exact = 1;
-
+
}
while (not_exact && !mpfr_can_round (t, Nt - 2, GMP_RNDN, GMP_RNDZ,
Nz + (rnd_mode == GMP_RNDN)));
diff --git a/init2.c b/init2.c
index 756c3e93c..80e59c005 100644
--- a/init2.c
+++ b/init2.c
@@ -1,6 +1,6 @@
/* mpfr_init2 -- initialize a floating-point number with given precision
-Copyright 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include <stdlib.h>
#include "mpfr-impl.h"
diff --git a/log.c b/log.c
index dd978fb01..8c1fc6622 100644
--- a/log.c
+++ b/log.c
@@ -1,6 +1,6 @@
/* mpfr_log -- natural logarithm of a floating-point number
-Copyright 1999, 2000, 2001, 2002, 2003 Free Software Foundation.
+Copyright 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -42,7 +41,7 @@ MA 02111-1307, USA. */
/* #define DEBUG */
int
-mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
+mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
{
int m, go_on, size, cancel, inexact = 0;
mp_prec_t p, q;
@@ -63,7 +62,7 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
/* check for infinity before zero */
else if (MPFR_IS_INF(a))
{
- if (MPFR_IS_NEG(a))
+ if (MPFR_IS_NEG(a))
/* log(-Inf) = NaN */
{
MPFR_SET_NAN(r);
@@ -75,7 +74,7 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
MPFR_SET_POS(r);
MPFR_RET(0);
}
- }
+ }
else if (MPFR_IS_ZERO(a))
{
MPFR_SET_INF(r);
@@ -85,7 +84,7 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
else
MPFR_ASSERTN(0);
}
-
+
/* If a is negative, the result is NaN */
if (MPFR_UNLIKELY( MPFR_IS_NEG(a) ))
{
@@ -103,7 +102,7 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
MPFR_CLEAR_FLAGS(r);
q=MPFR_PREC(r);
-
+
ref = mpfr_get_d1 (a) - 1.0;
if (ref<0)
ref=-ref;
@@ -127,11 +126,11 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
/* All the mpfr_t needed have a precision of p */
TMP_MARK(marker);
size=(p-1)/BITS_PER_MP_LIMB+1;
- MPFR_TMP_INIT(cstp, cst, p, size);
+ MPFR_TMP_INIT(cstp, cst, p, size);
MPFR_TMP_INIT(rapportp, rapport, p, size);
MPFR_TMP_INIT(agmp, agm, p, size);
- MPFR_TMP_INIT(tmp1p, tmp1, p, size);
- MPFR_TMP_INIT(tmp2p, tmp2, p, size);
+ MPFR_TMP_INIT(tmp1p, tmp1, p, size);
+ MPFR_TMP_INIT(tmp2p, tmp2, p, size);
MPFR_TMP_INIT(sp, s, p, size);
MPFR_TMP_INIT(mmp, mm, p, size);
@@ -146,8 +145,8 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
mpfr_div (tmp2, cst, tmp1, GMP_RNDN); /* pi/2*AG(1,4/s), err<=5ulps */
mpfr_const_log2 (cst, GMP_RNDN); /* compute log(2), err<=1ulp */
mpfr_mul(tmp1,cst,mm,GMP_RNDN); /* I compute m*log(2), err<=2ulps */
- cancel = MPFR_GET_EXP (tmp2);
- mpfr_sub(cst,tmp2,tmp1,GMP_RNDN); /* log(a), err<=7ulps+cancel */
+ cancel = MPFR_GET_EXP (tmp2);
+ mpfr_sub(cst,tmp2,tmp1,GMP_RNDN); /* log(a), err<=7ulps+cancel */
cancel -= MPFR_GET_EXP (cst);
#ifdef DEBUG
printf("canceled bits=%d\n", cancel);
@@ -173,9 +172,7 @@ mpfr_log (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
/* We clean */
TMP_FREE(marker);
-
+
}
return inexact; /* result is inexact */
}
-
-
diff --git a/log10.c b/log10.c
index ede7ab463..1a6b0b91a 100644
--- a/log10.c
+++ b/log10.c
@@ -1,6 +1,6 @@
/* mpfr_log10 -- logarithm in base 10.
-Copyright 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -69,7 +68,7 @@ mpfr_log10 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
else
MPFR_ASSERTN(0);
}
-
+
/* If a is negative, the result is NaN */
if (MPFR_UNLIKELY( MPFR_IS_NEG(a) ))
{
@@ -97,10 +96,10 @@ mpfr_log10 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
/* Declaration of the size variable */
mp_prec_t Nx = MPFR_PREC(a); /* Precision of input variable */
mp_prec_t Ny = MPFR_PREC(r); /* Precision of output variable */
-
+
mp_prec_t Nt; /* Precision of the intermediary variable */
long int err; /* Precision of error */
-
+
/* compute the precision of intermediary variable */
Nt = MAX(Nx, Ny);
/* the optimal number of bits : see algorithms.ps */
@@ -110,14 +109,14 @@ mpfr_log10 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
mpfr_init (t);
mpfr_init (tt);
-
+
/* First computation of log10 */
do {
/* reactualisation of the precision */
mpfr_set_prec (t, Nt);
- mpfr_set_prec (tt, Nt);
-
+ mpfr_set_prec (tt, Nt);
+
/* compute log10 */
mpfr_set_ui (t, 10, GMP_RNDN); /* 10 */
mpfr_log (t, t, GMP_RNDD); /* log(10) */
@@ -140,7 +139,7 @@ mpfr_log10 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
/* actualisation of the precision */
Nt += 10;
} while ((err < 0) || !ok);
-
+
inexact = mpfr_set (r, t, rnd_mode);
mpfr_clear (t);
diff --git a/log2.c b/log2.c
index fe0d23b78..37fcd2e75 100644
--- a/log2.c
+++ b/log2.c
@@ -1,6 +1,6 @@
/* mpfr_log2 -- log base 2
-Copyright 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -39,7 +38,7 @@ mpfr_log2 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
{
/* If a is NaN, the result is NaN */
if (MPFR_IS_NAN(a))
- {
+ {
MPFR_SET_NAN(r);
MPFR_RET_NAN;
}
@@ -68,7 +67,7 @@ mpfr_log2 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
else
MPFR_ASSERTN(0);
}
-
+
/* If a is negative, the result is NaN */
if (MPFR_UNLIKELY(MPFR_IS_NEG(a)))
{
@@ -96,10 +95,10 @@ mpfr_log2 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
/* Declaration of the size variable */
mp_prec_t Nx = MPFR_PREC(a); /* Precision of input variable */
mp_prec_t Ny = MPFR_PREC(r); /* Precision of input variable */
-
+
mp_prec_t Nt; /* Precision of the intermediary variable */
long int err; /* Precision of error */
-
+
/* compute the precision of intermediary variable */
Nt=MAX(Nx,Ny);
@@ -107,22 +106,22 @@ mpfr_log2 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
Nt=Nt+3+__gmpfr_ceil_log2(Nt);
/* initialise of intermediary variable */
- mpfr_init(t);
- mpfr_init(tt);
+ mpfr_init(t);
+ mpfr_init(tt);
+
-
/* First computation of log2 */
do
{
/* reactualisation of the precision */
- mpfr_set_prec(t,Nt);
- mpfr_set_prec(tt,Nt);
-
+ mpfr_set_prec(t,Nt);
+ mpfr_set_prec(tt,Nt);
+
/* compute log2 */
mpfr_const_log2(t,GMP_RNDD); /* log(2) */
mpfr_log(tt,a,GMP_RNDN); /* log(a) */
mpfr_div(t,tt,t,GMP_RNDN); /* log(a)/log(2) */
-
+
/* estimation of the error */
err=Nt-3;
@@ -131,7 +130,7 @@ mpfr_log2 (mpfr_ptr r, mpfr_srcptr a, mp_rnd_t rnd_mode)
}
while ((err < 0) || !mpfr_can_round (t, err, GMP_RNDN, GMP_RNDZ,
Ny + (rnd_mode == GMP_RNDN)));
-
+
inexact = mpfr_set (r, t, rnd_mode);
mpfr_clear (t);
diff --git a/reldiff.c b/reldiff.c
index 5335bc38e..df2a467bb 100644
--- a/reldiff.c
+++ b/reldiff.c
@@ -1,6 +1,6 @@
/* mpfr_reldiff -- compute relative difference of two floating-point numbers.
-Copyright 2000, 2001 Free Software Foundation, Inc.
+Copyright 2000, 2001, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,29 +19,28 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
#include "mpfr-impl.h"
/* reldiff(b, c) = abs(b-c)/b */
-void
+void
mpfr_reldiff (mpfr_ptr a, mpfr_srcptr b, mpfr_srcptr c, mp_rnd_t rnd_mode)
{
mpfr_t b_copy;
if (MPFR_IS_NAN(b) || MPFR_IS_NAN(c))
{ MPFR_CLEAR_FLAGS(a); MPFR_SET_NAN(a); return; }
- if (MPFR_IS_INF(b))
- {
+ if (MPFR_IS_INF(b))
+ {
if (MPFR_IS_INF(c) && (MPFR_SIGN(c) == MPFR_SIGN(b)))
{ MPFR_CLEAR_FLAGS(a); MPFR_SET_ZERO(a); return; }
else
{ MPFR_CLEAR_FLAGS(a); MPFR_SET_NAN(a); return; }
}
- if (MPFR_IS_INF(c))
+ if (MPFR_IS_INF(c))
{
MPFR_SET_SAME_SIGN(a, b);
MPFR_CLEAR_FLAGS(a);
@@ -62,9 +61,8 @@ mpfr_reldiff (mpfr_ptr a, mpfr_srcptr b, mpfr_srcptr c, mp_rnd_t rnd_mode)
mpfr_sub (a, b, c, rnd_mode);
mpfr_abs (a, a, rnd_mode); /* for compatibility with MPF */
mpfr_div (a, a, (a == b) ? b_copy : b, rnd_mode);
-
+
if (a == b)
mpfr_clear (b_copy);
}
}
-
diff --git a/save_expo.c b/save_expo.c
index 5f2a83651..33c4d70d3 100644
--- a/save_expo.c
+++ b/save_expo.c
@@ -1,6 +1,6 @@
/* Save/restore the minimum and maximum exponents.
-Copyright 2001, 2002 Free Software Foundation, Inc.
+Copyright 2001, 2002, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,8 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
-#include <stdlib.h>
#include "gmp.h"
#include "mpfr.h"
#include "mpfr-impl.h"
@@ -41,12 +39,9 @@ mpfr_save_emin_emax (void)
__gmpfr_emin = MPFR_EMIN_MIN;
__gmpfr_emax = MPFR_EMAX_MAX;
}
- else if (save_ctr == 0)
+ else
{
- fprintf(stderr,
- "Error: Too many consecutive calls to mpfr_save_emin_emax!\n"
- "Probably a bug.\n");
- exit(1);
+ MPFR_ASSERTN (save_ctr != 0);
}
}
diff --git a/set_f.c b/set_f.c
index 65fe42110..39f628af6 100644
--- a/set_f.c
+++ b/set_f.c
@@ -1,6 +1,6 @@
/* mpfr_set_f -- set a MPFR number from a GNU MPF number
-Copyright 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "longlong.h"
@@ -50,7 +49,7 @@ mpfr_set_f (mpfr_ptr y, mpf_srcptr x, mp_rnd_t rnd_mode)
return 0; /* 0 is exact */
}
- count_leading_zeros(cnt, mx[sx - 1]);
+ count_leading_zeros(cnt, mx[sx - 1]);
if (sy <= sx) /* we may have to round even when sy = sx */
{
@@ -69,14 +68,14 @@ mpfr_set_f (mpfr_ptr y, mpf_srcptr x, mp_rnd_t rnd_mode)
else
{
if (cnt)
- mpn_lshift(my + sy - sx, mx, sx, cnt);
+ mpn_lshift(my + sy - sx, mx, sx, cnt);
else
- MPN_COPY(my + sy - sx, mx, sy);
+ MPN_COPY(my + sy - sx, mx, sy);
MPN_ZERO(my, sy - sx);
/* no rounding necessary, since y has a larger mantissa */
inexact = 0;
}
-
+
MPFR_SET_EXP(y, EXP(x) * BITS_PER_MP_LIMB - cnt);
return inexact;
diff --git a/set_prc_raw.c b/set_prc_raw.c
index 2bc8e1d45..aa1eb5afe 100644
--- a/set_prc_raw.c
+++ b/set_prc_raw.c
@@ -1,6 +1,6 @@
/* mpfr_set_prec_raw -- reset the precision of a floating-point number
-Copyright 2000, 2001 Free Software Foundation, Inc.
+Copyright 2000, 2001, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,8 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
-#include <stdlib.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -29,13 +27,7 @@ MA 02111-1307, USA. */
void
mpfr_set_prec_raw (mpfr_ptr x, mpfr_prec_t p)
{
- MPFR_ASSERTN(p >= MPFR_PREC_MIN && p <= MPFR_PREC_MAX);
-
- if (p > (mpfr_prec_t) MPFR_GET_ALLOC_SIZE(x) * BITS_PER_MP_LIMB)
- {
- fprintf (stderr, "*** precision too large for allocated space\n");
- exit (1);
- }
-
+ MPFR_ASSERTN (p >= MPFR_PREC_MIN && p <= MPFR_PREC_MAX);
+ MPFR_ASSERTN (p <= (mpfr_prec_t) MPFR_GET_ALLOC_SIZE(x) * BITS_PER_MP_LIMB);
MPFR_PREC(x) = p;
}
diff --git a/set_prec.c b/set_prec.c
index 113da143b..7a01cbd8a 100644
--- a/set_prec.c
+++ b/set_prec.c
@@ -1,6 +1,6 @@
/* mpfr_set_prec -- reset the precision of a floating-point number
-Copyright 1999, 2001, 2002 Free Software Foundation, Inc.
+Copyright 1999, 2001, 2002, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "mpfr.h"
@@ -34,7 +33,7 @@ mpfr_set_prec (mpfr_ptr x, mpfr_prec_t p)
/* first, check if p is correct */
MPFR_ASSERTN(p >= MPFR_PREC_MIN && p <= MPFR_PREC_MAX);
- /* Calculate the new number of limbs */
+ /* Calculate the new number of limbs */
xsize = (p - 1) / BITS_PER_MP_LIMB + 1;
/* Realloc only if the new size is greater than the old */
diff --git a/set_str.c b/set_str.c
index fcd43505c..4b49af0a5 100644
--- a/set_str.c
+++ b/set_str.c
@@ -1,6 +1,6 @@
/* mpfr_set_str -- set a floating-point number from a string
-Copyright 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>
@@ -144,7 +143,7 @@ mpfr_set_str (mpfr_t x, const char *str, int base, mp_rnd_t rnd)
if (**endptr != '\0')
{
res = -1; /* invalid input: garbage after exponent */
- goto end;
+ goto end;
}
break;
}
@@ -160,18 +159,18 @@ mpfr_set_str (mpfr_t x, const char *str, int base, mp_rnd_t rnd)
break;
}
else if (*str == '.')
- {
+ {
point = 1;
continue; /* go to next character */
}
else if ((c = digit_value_in_base (*str, base)) != -1)
*mant_s = c; /* valid character */
- else
+ else
{
res = -1; /* invalid input */
goto end;
}
- if (!point)
+ if (!point)
exp_s ++;
mant_s ++;
}
@@ -209,7 +208,7 @@ mpfr_set_str (mpfr_t x, const char *str, int base, mp_rnd_t rnd)
/* remember that y - n is allocated for n limbs */
/* required precision for str */
- pr = (mp_exp_t) __gmpfr_ceil
+ pr = (mp_exp_t) __gmpfr_ceil
(((double) (n * BITS_PER_MP_LIMB) - 1.0) * log_b2[base-2]) + 1;
/* check is there are enough digits in str */
@@ -278,7 +277,7 @@ mpfr_set_str (mpfr_t x, const char *str, int base, mp_rnd_t rnd)
/* compute the exponent of y */
exp_y += exp_z + n * BITS_PER_MP_LIMB;
-
+
/* normalize result */
if ((result[2 * n - 1] & MPFR_LIMB_HIGHBIT) == 0)
{
@@ -355,7 +354,7 @@ mpfr_set_str (mpfr_t x, const char *str, int base, mp_rnd_t rnd)
negative, MPFR_PREC(x), rnd ))
{
/* overflaw when rounding y */
- MPFR_MANT(x)[MPFR_LIMB_SIZE(x) - 1]
+ MPFR_MANT(x)[MPFR_LIMB_SIZE(x) - 1]
= MPFR_LIMB_HIGHBIT;
exp_y ++;
}
diff --git a/ui_div.c b/ui_div.c
index 3150f371c..3a45d0aea 100644
--- a/ui_div.c
+++ b/ui_div.c
@@ -1,6 +1,6 @@
/* mpfr_ui_div -- divide a machine integer by a floating-point number
-Copyright 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
+Copyright 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
This file is part of the MPFR Library.
@@ -19,7 +19,6 @@ along with the MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
-#include <stdio.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "longlong.h"
@@ -59,7 +58,7 @@ mpfr_ui_div (mpfr_ptr y, unsigned long int u, mpfr_srcptr x, mp_rnd_t rnd_mode)
{
/* 0 / 0 */
MPFR_SET_NAN(y);
- MPFR_RET_NAN;
+ MPFR_RET_NAN;
}
}
else
diff --git a/zeta.c b/zeta.c
index d5c706492..143164b09 100644
--- a/zeta.c
+++ b/zeta.c
@@ -1,6 +1,6 @@
/* mpfr_zeta -- compute the Riemann Zeta function
-Copyright 2003 Free Software Foundation.
+Copyright 2003, 2004 Free Software Foundation.
Contributed by Jean-Luc Re'my and the Spaces project, INRIA Lorraine.
This file is part of the MPFR Library.
@@ -22,7 +22,6 @@ MA 02111-1307, USA. */
/* #define DEBUG */
-#include <stdio.h>
#include <stdlib.h>
#include "gmp.h"
#include "gmp-impl.h"