Copyright question.

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Copyright question.

Dwayne Grant McConnell-2

So for the "IBM extended format" 128-bit long double implementation I
started with some existing files, say
libc/sysdeps/ieee754/ldbl-128/e_acosl.c, and modified them for the new
format. What should the copyright look like now? Should I add a third
copyright notice corresponding to the FSF one? Should I edit the long
double copyright and if so to what extent? Should I just leave it alone? I
was assuming that my name would not be included because the copyright has
been assigned to FSF but now that I see Stephen L. Moshier's name below
I'm unsure. The original file is enclosed below.

Also when borrowing/modifying code like this is it better to follow the
existing indentation scheme or to reformat according to GNU Coding
Standards?

Thanks,
Dwayne

--
Dwayne Grant McConnell <[hidden email]>
Lotus Notes Mail: Dwayne McConnell [Mail]/Austin/IBM@IBMUS
Lotus Notes Calendar: Dwayne McConnell [Calendar]/Austin/IBM@IBMUS

----
/*
 * ====================================================
 * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
 *
 * Developed at SunPro, a Sun Microsystems, Inc. business.
 * Permission to use, copy, modify, and distribute this
 * software is freely granted, provided that this notice
 * is preserved.
 * ====================================================
 */

/*
   Long double expansions are
   Copyright (C) 2001 Stephen L. Moshier <[hidden email]>
   and are incorporated herein by permission of the author.  The author
   reserves the right to distribute this material elsewhere under different
   copying permissions.  These modifications are distributed here under
   the following terms:

    This library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Lesser General Public
    License as published by the Free Software Foundation; either
    version 2.1 of the License, or (at your option) any later version.

    This library is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
    Lesser General Public License for more details.

    You should have received a copy of the GNU Lesser General Public
    License along with this library; if not, write to the Free Software
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA */

/* __ieee754_acosl(x)
 * Method :
 *      acos(x)  = pi/2 - asin(x)
 *      acos(-x) = pi/2 + asin(x)
 * For |x| <= 0.375
 *      acos(x) = pi/2 - asin(x)
 * Between .375 and .5 the approximation is
 *      acos(0.4375 + x) = acos(0.4375) + x P(x) / Q(x)
 * Between .5 and .625 the approximation is
 *      acos(0.5625 + x) = acos(0.5625) + x rS(x) / sS(x)
 * For x > 0.625,
 *      acos(x) = 2 asin(sqrt((1-x)/2))
 *      computed with an extended precision square root in the leading term.
 * For x < -0.625
 *      acos(x) = pi - 2 asin(sqrt((1-|x|)/2))
 *
 * Special cases:
 *      if x is NaN, return x itself;
 *      if |x|>1, return NaN with invalid signal.
 *
 * Functions needed: __ieee754_sqrtl.
 */

#include "math.h"
#include "math_private.h"

#ifdef __STDC__
static const long double
#else
static long double
#endif
  one = 1.0L,
  pio2_hi = 1.5707963267948966192313216916397514420986L,
  pio2_lo = 4.3359050650618905123985220130216759843812E-35L,

  /* acos(0.5625 + x) = acos(0.5625) + x rS(x) / sS(x)
     -0.0625 <= x <= 0.0625
     peak relative error 3.3e-35  */

  rS0 =  5.619049346208901520945464704848780243887E0L,
  rS1 = -4.460504162777731472539175700169871920352E1L,
  rS2 =  1.317669505315409261479577040530751477488E2L,
  rS3 = -1.626532582423661989632442410808596009227E2L,
  rS4 =  3.144806644195158614904369445440583873264E1L,
  rS5 =  9.806674443470740708765165604769099559553E1L,
  rS6 = -5.708468492052010816555762842394927806920E1L,
  rS7 = -1.396540499232262112248553357962639431922E1L,
  rS8 =  1.126243289311910363001762058295832610344E1L,
  rS9 =  4.956179821329901954211277873774472383512E-1L,
  rS10 = -3.313227657082367169241333738391762525780E-1L,

  sS0 = -4.645814742084009935700221277307007679325E0L,
  sS1 =  3.879074822457694323970438316317961918430E1L,
  sS2 = -1.221986588013474694623973554726201001066E2L,
  sS3 =  1.658821150347718105012079876756201905822E2L,
  sS4 = -4.804379630977558197953176474426239748977E1L,
  sS5 = -1.004296417397316948114344573811562952793E2L,
  sS6 =  7.530281592861320234941101403870010111138E1L,
  sS7 =  1.270735595411673647119592092304357226607E1L,
  sS8 = -1.815144839646376500705105967064792930282E1L,
  sS9 = -7.821597334910963922204235247786840828217E-2L,
  /* 1.000000000000000000000000000000000000000E0 */

  acosr5625 = 9.7338991014954640492751132535550279812151E-1L,
  pimacosr5625 = 2.1682027434402468335351320579240000860757E0L,

  /* acos(0.4375 + x) = acos(0.4375) + x rS(x) / sS(x)
     -0.0625 <= x <= 0.0625
     peak relative error 2.1e-35  */

  P0 =  2.177690192235413635229046633751390484892E0L,
  P1 = -2.848698225706605746657192566166142909573E1L,
  P2 =  1.040076477655245590871244795403659880304E2L,
  P3 = -1.400087608918906358323551402881238180553E2L,
  P4 =  2.221047917671449176051896400503615543757E1L,
  P5 =  9.643714856395587663736110523917499638702E1L,
  P6 = -5.158406639829833829027457284942389079196E1L,
  P7 = -1.578651828337585944715290382181219741813E1L,
  P8 =  1.093632715903802870546857764647931045906E1L,
  P9 =  5.448925479898460003048760932274085300103E-1L,
  P10 = -3.315886001095605268470690485170092986337E-1L,
  Q0 = -1.958219113487162405143608843774587557016E0L,
  Q1 =  2.614577866876185080678907676023269360520E1L,
  Q2 = -9.990858606464150981009763389881793660938E1L,
  Q3 =  1.443958741356995763628660823395334281596E2L,
  Q4 = -3.206441012484232867657763518369723873129E1L,
  Q5 = -1.048560885341833443564920145642588991492E2L,
  Q6 =  6.745883931909770880159915641984874746358E1L,
  Q7 =  1.806809656342804436118449982647641392951E1L,
  Q8 = -1.770150690652438294290020775359580915464E1L,
  Q9 = -5.659156469628629327045433069052560211164E-1L,
  /* 1.000000000000000000000000000000000000000E0 */

  acosr4375 = 1.1179797320499710475919903296900511518755E0L,
  pimacosr4375 = 2.0236129215398221908706530535894517323217E0L,

  /* asin(x) = x + x^3 pS(x^2) / qS(x^2)
     0 <= x <= 0.5
     peak relative error 1.9e-35  */
  pS0 = -8.358099012470680544198472400254596543711E2L,
  pS1 =  3.674973957689619490312782828051860366493E3L,
  pS2 = -6.730729094812979665807581609853656623219E3L,
  pS3 =  6.643843795209060298375552684423454077633E3L,
  pS4 = -3.817341990928606692235481812252049415993E3L,
  pS5 =  1.284635388402653715636722822195716476156E3L,
  pS6 = -2.410736125231549204856567737329112037867E2L,
  pS7 =  2.219191969382402856557594215833622156220E1L,
  pS8 = -7.249056260830627156600112195061001036533E-1L,
  pS9 =  1.055923570937755300061509030361395604448E-3L,

  qS0 = -5.014859407482408326519083440151745519205E3L,
  qS1 =  2.430653047950480068881028451580393430537E4L,
  qS2 = -4.997904737193653607449250593976069726962E4L,
  qS3 =  5.675712336110456923807959930107347511086E4L,
  qS4 = -3.881523118339661268482937768522572588022E4L,
  qS5 =  1.634202194895541569749717032234510811216E4L,
  qS6 = -4.151452662440709301601820849901296953752E3L,
  qS7 =  5.956050864057192019085175976175695342168E2L,
  qS8 = -4.175375777334867025769346564600396877176E1L;
  /* 1.000000000000000000000000000000000000000E0 */

#ifdef __STDC__
long double
__ieee754_acosl (long double x)
#else
long double
__ieee754_acosl (x)
     long double x;
#endif
{
  long double z, r, w, p, q, s, t, f2;
  int32_t ix, sign;
  ieee854_long_double_shape_type u;

  u.value = x;
  sign = u.parts32.w0;
  ix = sign & 0x7fffffff;
  u.parts32.w0 = ix; /* |x| */
  if (ix >= 0x3fff0000) /* |x| >= 1 */
    {
      if (ix == 0x3fff0000
          && (u.parts32.w1 | u.parts32.w2 | u.parts32.w3) == 0)
        { /* |x| == 1 */
          if ((sign & 0x80000000) == 0)
            return 0.0; /* acos(1) = 0  */
          else
            return (2.0 * pio2_hi) + (2.0 * pio2_lo); /* acos(-1)= pi */
        }
      return (x - x) / (x - x); /* acos(|x| > 1) is NaN */
    }
  else if (ix < 0x3ffe0000) /* |x| < 0.5 */
    {
      if (ix < 0x3fc60000) /* |x| < 2**-57 */
        return pio2_hi + pio2_lo;
      if (ix < 0x3ffde000) /* |x| < .4375 */
        {
          /* Arcsine of x.  */
          z = x * x;
          p = (((((((((pS9 * z
                       + pS8) * z
                      + pS7) * z
                     + pS6) * z
                    + pS5) * z
                   + pS4) * z
                  + pS3) * z
                 + pS2) * z
                + pS1) * z
               + pS0) * z;
          q = (((((((( z
                       + qS8) * z
                     + qS7) * z
                    + qS6) * z
                   + qS5) * z
                  + qS4) * z
                 + qS3) * z
                + qS2) * z
               + qS1) * z
            + qS0;
          r = x + x * p / q;
          z = pio2_hi - (r - pio2_lo);
          return z;
        }
      /* .4375 <= |x| < .5 */
      t = u.value - 0.4375L;
      p = ((((((((((P10 * t
                    + P9) * t
                   + P8) * t
                  + P7) * t
                 + P6) * t
                + P5) * t
               + P4) * t
              + P3) * t
             + P2) * t
            + P1) * t
           + P0) * t;

      q = (((((((((t
                   + Q9) * t
                  + Q8) * t
                 + Q7) * t
                + Q6) * t
               + Q5) * t
              + Q4) * t
             + Q3) * t
            + Q2) * t
           + Q1) * t
        + Q0;
      r = p / q;
      if (sign & 0x80000000)
        r = pimacosr4375 - r;
      else
        r = acosr4375 + r;
      return r;
    }
  else if (ix < 0x3ffe4000) /* |x| < 0.625 */
    {
      t = u.value - 0.5625L;
      p = ((((((((((rS10 * t
                    + rS9) * t
                   + rS8) * t
                  + rS7) * t
                 + rS6) * t
                + rS5) * t
               + rS4) * t
              + rS3) * t
             + rS2) * t
            + rS1) * t
           + rS0) * t;

      q = (((((((((t
                   + sS9) * t
                  + sS8) * t
                 + sS7) * t
                + sS6) * t
               + sS5) * t
              + sS4) * t
             + sS3) * t
            + sS2) * t
           + sS1) * t
        + sS0;
      if (sign & 0x80000000)
        r = pimacosr5625 - p / q;
      else
        r = acosr5625 + p / q;
      return r;
    }
  else
    { /* |x| >= .625 */
      z = (one - u.value) * 0.5;
      s = __ieee754_sqrtl (z);
      /* Compute an extended precision square root from
         the Newton iteration  s -> 0.5 * (s + z / s).
         The change w from s to the improved value is
            w = 0.5 * (s + z / s) - s  = (s^2 + z)/2s - s = (z - s^2)/2s.
          Express s = f1 + f2 where f1 * f1 is exactly representable.
          w = (z - s^2)/2s = (z - f1^2 - 2 f1 f2 - f2^2)/2s .
          s + w has extended precision.  */
      u.value = s;
      u.parts32.w2 = 0;
      u.parts32.w3 = 0;
      f2 = s - u.value;
      w = z - u.value * u.value;
      w = w - 2.0 * u.value * f2;
      w = w - f2 * f2;
      w = w / (2.0 * s);
      /* Arcsine of s.  */
      p = (((((((((pS9 * z
                   + pS8) * z
                  + pS7) * z
                 + pS6) * z
                + pS5) * z
               + pS4) * z
              + pS3) * z
             + pS2) * z
            + pS1) * z
           + pS0) * z;
      q = (((((((( z
                   + qS8) * z
                 + qS7) * z
                + qS6) * z
               + qS5) * z
              + qS4) * z
             + qS3) * z
            + qS2) * z
           + qS1) * z
        + qS0;
      r = s + (w + s * p / q);

      if (sign & 0x80000000)
        w = pio2_hi + (pio2_lo - r);
      else
        w = r;
      return 2.0 * w;
    }
}
Reply | Threaded
Open this post in threaded view
|

Re: Copyright question.

Roland McGrath
When you modify any libc source file, you must leave any non-FSF copyright
notices and permission text completely unmodified.  When you copy any
significant part of one libc source file into into another source file, you
must copy any non-FSF copyright notices and permission text verbatim into
the new file.  When there is an FSF copyright notice, you should add the
current year to its list of years.  If there is no FSF copyright notice,
and you are adding significant code beyond what was copied, then you should
add a new FSF copyright notice that looks like the ones you see in all the
other libc files that have no non-FSF copyrights.

Sun and Stephen Moshier both have retained their copyright ownership of the
libm code we have in libc.  That's why their names will always be there.
The copyrights for your contributions are assigned to the FSF, so if you
are adding anything copyrightable (i.e. significant, which means anything
more than a few lines) then the only copyright being asserted is the FSF's.

Many people put a "Written by" or "Contributed by" line right after the
copyright line (or a "Converted to foobar support by" or whatever makes
sense for code that's been copied and extended).  That is not a copyright
and has no legal meaning.  The copyright notices (such as Moshier's) cannot
be removed or changed legally.  The notes are just text in the file that we
could choose to change or remove, and we just choose to leave them in there
since that's what those contributors seemed to want went they gave us the
code.  You're welcome to put such notes in your submissions if you like.
It's up to you.  (Many people write them; I never do, but that's just
because the last time I put my name or email address near the top of a
source file, I was still getting highly confused email from novice
programmers in Korea fifteen years later.)

As for reformatting, usually it's better avoided.  If the code you've
copied is still substantially similar to the old code, then you'll be doing
yourself a favor not to perturb it unnecessarily even with whitespace
changes, so that diffs between the two, and the diffs for future changes to
the code you copied from, are easier to read and see how to apply to your
code.  That's why we didn't reformat all the Sun libm code when we took it
in.  If you are really rewriting all the text so the similarity is lost to
diff anyway, then use GNU conventions.  (The one file you posted is already
indented in the GNU style.)


Thanks,
Roland