1 2004-07-01, version 1.1.8
2 =========================
7 * Fix bug in log(a,b) when b is 1/n.
9 * Fix crash in shared library on AMD64.
14 * Fixes for some language conformance issues.
17 2004-05-02, version 1.1.7
18 =========================
20 Implementation changes
21 ----------------------
23 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
25 * Workaround GCC 3.[123] compiler bug.
27 * Parallel build should work reliably now.
30 2004-01-01, version 1.1.6
31 =========================
37 `sintL ldegree(const cl_UP& x)'.
38 (Suggested by Munagala Ramanath.)
40 Implementation changes
41 ----------------------
43 * Fixed a bug in polynomial subtraction.
44 (Reported by Munagala Ramanath.)
49 * Provide a package control file cln.pc.
51 * Updated infrastructure to autoconf-2.57.
53 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
56 2002-05-28, version 1.1.5
57 =========================
62 * Fix compilation issues with GCC-3.1.
65 2002-01-04, version 1.1.4
66 =========================
71 * Fix compilation issues with GMP-4.0 and on IBM s390.
73 * Updated infrastructure to autoconf-2.52.
76 2001-11-05, version 1.1.3
77 =========================
79 Implementation changes
80 ----------------------
82 * Fix a computation error of sin and sinh for small arguments.
83 (Reported by Christopher Kennedy.)
85 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
87 * -DNO_ASM works properly on Sparc.
90 2001-07-25, version 1.1.2
92 Implementation changes
93 ----------------------
95 * Minor cleanups and portability issues.
96 (No need to upgrade if you have been happy so far.)
99 2001-05-31, version 1.1.1
100 =========================
102 Implementation changes
103 ----------------------
105 * Minor cleanups for GCC 3.0 and libstdc++-v3.
107 * Fixes several little bugs and portability issues.
110 2000-12-14, version 1.1.0
111 =========================
116 * ISO/IEC 14882 fashion adjustments:
117 Put everything into namespace cln. All fundamental data types still
118 carry their old names. Other non-macro identifiers are now written as
119 cln::foo instead of cl_foo, except where the cl_ comes from a data type
120 (as in cl_I_to_int()). Headers are installed into a separate
121 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
122 must be manually ported to the new style. We apologize for the
125 Implementation changes
126 ----------------------
128 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
129 Configuring with --with-gmp now relies on an installed libgmp library
130 version 3 or above. We recommend libgmp-3.1 or newer.
132 * Adjusted several break-even points to make better use of a faster libgmp and
133 better match present-day CPUs.
135 * Fix several errors for architectures with 64-bit wide words.
136 (Others than Alpha, which worked already.)
138 * Fix compilation errors with current GCC-3.0 snapshots.
143 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
147 2000-01-13, version 1.0.3
148 =========================
154 `cl_I doublefactorial (uintL n)'.
156 Implementation changes
157 ----------------------
159 * Fix several little configuration errors.
161 * Fix some compilation errors with gcc-2.96 prereleases.
164 1999-09-07, version 1.0.2
165 =========================
171 `cl_I numerator (const cl_RA&)'
172 `cl_I denominator (const cl_RA&)'.
173 Suggested by Richard Kreckel and Sylvain Pion.
175 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
176 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
177 Suggested by Richard Kreckel.
179 Implementation changes
180 ----------------------
182 * Fix an endless loop when either of the functions
183 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
184 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
185 was called with an integer argument. Thanks to Richard Kreckel.
187 * Fix a bug in the addition and subtraction of rational numbers which could
188 cause unnormalized rational numbers like 3/1 to be created.
191 1999-06-09, version 1.0.1
192 =========================
197 * Speed up the functions `logand', `lognand', `logandc2'.
199 Implementation changes
200 ----------------------
202 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
204 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
205 and SPARC with LiDIA.
207 * Fix compilation errors on Win32. Thanks to Mumit Khan.
210 1999-01-12, version 1.0
211 =======================
216 * New include files, containing I/O functions:
220 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
223 The file <cl_io.h> now contains only I/O functions for characters and
226 * To access the base ring of a univariate polynomial ring, now use
227 `R->basering()' instead of `R->basering'.
229 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
232 * Implement `abs' for the class cl_RA.
234 * Replaced `read_number' with specialized functions `read_complex',
235 `read_real', `read_float', `read_rational', `read_integer'.
237 * Replaced the functions
238 `void fprint (cl_ostream stream, int x)'
239 `void fprint (cl_ostream stream, unsigned int x)'
240 `void fprint (cl_ostream stream, long x)'
241 `void fprint (cl_ostream stream, unsigned long x)'
243 `void fprintdecimal (cl_ostream stream, int x)'
244 `void fprintdecimal (cl_ostream stream, unsigned int x)'
245 `void fprintdecimal (cl_ostream stream, long x)'
246 `void fprintdecimal (cl_ostream stream, unsigned long x)'
251 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
252 now normalizes the resulting u and v to be of minimum absolute value.
253 (Suggested by the LiDIA group.)
255 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
256 the resulting float depending on the number of significant digits present
259 * Speed up the functions
260 `cl_R operator* (const cl_RA&, const cl_LF&)'
261 `cl_R operator* (const cl_LF&, const cl_RA&)'
262 `cl_R operator/ (const cl_RA&, const cl_LF&)'
263 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
265 Implementation changes
266 ----------------------
268 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
270 (Reported by Thomas Papanikolaou.)
272 * Fixed a memory leak in the long-float truncate function.
273 (Reported by Keith Briggs.)
275 * Fixed a memory leak in the modular integer exponentiation.
277 * Reduced the stack space consumption. The default stack size (typically
278 8 MB on Unix) should now be sufficient in order to run all CLN programs.
280 * Portability fixes for m68k platforms.
281 (Reported and fixed by Roman Hodek.)
286 * Restructured the sources. The subpackages
289 - base + integer + rational
290 - base + integer + rational + float
291 - base + integer + rational + float + real
292 - base + integer + rational + float + real + complex
293 - base + integer + modinteger
294 are now self-contained.
296 * The configuration script can be called with option `--enable-shared',
297 to build CLN as a shared library.
299 * Updated the LiDIA interface.
302 1998-05-07, version 0.98
303 ========================
305 Implementation changes
306 ----------------------
308 * Portability fixes for 64-bit platforms like DEC Alpha.
309 (Reported by John Cremona and Thomas Papanikolaou.)
318 * Portability fixes for new ISO/ANSI C++ compilers.
320 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
327 Implementation changes
328 ----------------------
330 * Portability fix for i686 platforms.
331 (Reported by Michael Pfeifer.)
341 <cl_ring.h>, declaring general rings,
342 <cl_GV.h>, <cl_SV.h>, defining vectors,
343 <cl_univpoly.h>, defining univariate polynomial rings.
348 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
350 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
352 Implementation changes
353 ----------------------
355 * Portability fix for 64-bit platforms like DEC Alpha:
356 Fixed a bug: endless loop in `sqrt' for large precision.
357 (Reported by Blake Jones.)
359 * Portability fix for RS/6000 platforms.
368 * Speed up the computation of the elementary transcendental functions:
369 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
370 log (for 100 digits) by about 25%,
371 sin, cos (for 1000 or 10000 digits) by about 20%,
372 asin, acos, atan (for 1000 digits) by about 25%.
374 Implementation changes
375 ----------------------
377 * Portability fix for SunOS 4 and other BSD platforms.
386 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
388 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
389 for moderate precision (< 11370 digits) were correct.
390 (Reported by Thomas Papanikolaou.)
392 Implementation changes
393 ----------------------
395 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
396 (g++ 2.8.0 or newer on HPPA).
398 * New method `debug_print()', for pretty printing of CLN objects, intended
399 to be called from the debugger.
408 * New functions `float cl_float_approx (const cl_R&)',
409 `double cl_double_approx (const cl_R&)'
410 for converting a real number to a C `float' or `double'.
415 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
417 Implementation changes
418 ----------------------
420 * New function `cl_print', for pretty printing of CLN objects, intended to be
421 called from the debugger.
423 * Portability fixes for 64-bit platforms like DEC Alpha.
433 <cl_string.h>, defining strings,
434 <cl_symbol.h>, defining symbols,
435 <cl_proplist.h>, defining property lists,
436 <cl_condition.h>, defining conditions/exceptions.
438 * New functions `cl_F cl_catalanconst ()',
439 `cl_F cl_catalanconst (const cl_F&)',
440 `cl_F cl_catalanconst (cl_float_format_t)'
441 which return Catalan's constant.
442 (Suggested by Thomas Papanikolaou.)
444 * New functions `cl_F cl_zeta (int s)',
445 `cl_F cl_zeta (int s, const cl_F&)',
446 `cl_F cl_zeta (int s, cl_float_format_t)'
447 which return the Riemann zeta function at an integral point s>1.
449 * New functions `cl_F cl_exp1 ()',
450 `cl_F cl_exp1 (const cl_F&)',
451 `cl_F cl_exp1 (cl_float_format_t)'
452 which return e = exp(1).
454 * New function `cl_I binomial (uintL n, uintL k)' which returns the
455 binomial coefficient (n choose k).
457 * New functions `int cl_I_to_int (const cl_I&)',
458 `unsigned int cl_I_to_uint (const cl_I&)',
459 `long cl_I_to_long (const cl_I&)',
460 `unsigned long cl_I_to_ulong (const cl_I&)'
461 for converting an integer to a C `int' or `long'.
463 * New functions `float cl_float_approx (const cl_I&)',
464 `float cl_float_approx (const cl_RA&)',
465 `double cl_double_approx (const cl_I&)',
466 `double cl_double_approx (const cl_RA&)'
467 for converting a rational number to a C `float' or `double'.
469 Implementation changes
470 ----------------------
472 * Moved the sources to subdirectories.
473 (Suggested by Jörg Arndt.)
478 * New benchmark for better comparing LiDIA, Pari and CLN.
480 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
481 (Suggested by Thomas Papanikolaou.)
483 * Added an ILOG Talk interface, for interactive use of the library.
492 * New include file <cl_modinteger.h>, defining modular integer rings.
494 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
495 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
500 * Speed up square root of large numbers by use of Newton's algorithm.
502 * Speed up multiplication and division of large long-floats by small integers.
504 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
505 asin, acos, atan, asinh, acosh, atanh.
506 All of them are now computed with asymptotically fast algorithms, of
507 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
509 * Fixed several bugs in the transcendental functions routines. In particular,
510 the `log' function went into an endless loop.
512 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
513 correctly for negative integers.
522 * New include file <cl_numtheory.h>, containing number theoretic functions,
523 for now only the Jacobi symbol.
528 * Speed up squaring of large numbers by 30%.
530 * Speed up division of large numbers by use of Newton's algorithm. The
531 improvement is noticeable for numbers with at least about 1000 words =
532 32000 bits = 10000 decimal digits.
534 * Speed up the binary-to-decimal conversion of large numbers by use of
535 a divide-and-conquer method. The improvement is noticeable for numbers
536 of at least 1250 words = 40000 bits = 12000 decimal digits.
538 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
539 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
541 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
542 didn't work correctly if both arguments were bignums.
544 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
545 as `mod(-n*m,m)' (n>0), returned m instead of 0.
547 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
548 values for arguments x = 2^n, n >= 29.
550 Implementation changes
551 ----------------------
553 * Speed up CLN by using the low-level addition/subtraction, shift and
554 multiplication loops from GNU gmp-2.0.2.
556 * Fixed a severe reference counting bug in the constructors
557 `type::type (const long)' and `type::type (const unsigned long)'
558 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
561 * CLN's word sequences can be chosen to be stored big-endian in memory
562 (like it was up to now) or little-endian (needed for interfacing to GMP).
571 * New include file <cl_timing.h>, contains timing facilities.
576 * Speed up the multiplication of large numbers by use of an FFT based
577 algorithm (Schönhage-Strassen). The improvement is noticeable when both
578 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
580 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
581 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
582 of 1.5. Simple trick.
584 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
585 trick from Henri Cohen.
587 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
588 which occurred when one of the arguments is zero.
590 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
592 Implementation changes
593 ----------------------
595 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
596 (Reported by Tito Flagella.)
598 * Fixed linking error when compiling without -O.
599 (Reported by Tito Flagella.)
601 * Fixed syntax error in <cl_input.h>.
602 (Reported by Tito Flagella.)
604 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
605 (Reported by Marcus Daniels.)
607 * More portability fixes.
609 * Configuration now supports the --srcdir option.
610 (Reported by Marcus Daniels.)
616 * Compiles well on Sparc, using gcc.
618 * Mail to beta testers.
624 * Compiles well on Linux, using gcc.