1 2009-06-30, version 1.3.0
2 =========================
7 * Use the GNU autotools as build system.
9 * Implemented a more portable module dependency mechanism.
12 2008-04-05, version 1.2.2
13 =========================
15 Implementation changes
16 ----------------------
18 * Re-establish CLN-1.2.0 ABI and fix ARM build, both inadvertently broken in
22 2008-03-24, version 1.2.1
23 =========================
25 Implementation changes
26 ----------------------
28 * Fixed some bugs in the output of numbers larger than 2^32 decimal digits.
33 * Modifying C/C++ operators like +=, ++, etc. are now enabled by default.
36 2008-01-19, version 1.2.0
37 =========================
42 * Save big amounts of memory in computation of some functions and constants by:
43 - Avoiding pre-computation of series terms and instead computing them in a
45 - Avoiding computation with excess precision in binary splitting algorithm
46 by coercion of intermediate integer values to floating-point as early as
49 Implementation changes
50 ----------------------
52 * Added support for huge numbers:
53 - intC used for all counter loops.
54 - intE is now a 64-bit type on all 64-bit platforms and even on selected
57 * CLN now uses C++ exceptions for error handling. The cl_abort() hook is not
58 supported any more. Please refer to the documentation to learn about existing
61 * Fixed a bug on i386 where comparing Z/2Z ring zeros returnd random results.
63 * Removed cl_boolean. Use built-in C++ bool instead.
68 * Dropped the cln-config script. Please use pkg-config instead.
70 * Updated infrastructure to that of libtool-1.5.24.
72 * Changed encoding of all files to UTF-8.
74 * Fix compilation issues with GCC-4.3.
76 * Fix linking issues on platforms that don't feature weak symbols (like win32).
79 2006-08-08, version 1.1.13
80 ==========================
82 * Compilation fixes for 64-bit brokenness introduced in last release.
85 2006-08-06, version 1.1.12
86 ==========================
88 Implementation changes
89 ----------------------
91 * Fix rare assertion when printing quite large floats.
96 * Compilation fixes for several platforms: *BSD, Intel Mac, and MinGW.
99 2005-11-23, version 1.1.11
100 ==========================
105 * Considerably improved performance of number input.
108 2005-10-22, version 1.1.10
109 ==========================
111 Implementation changes
112 ----------------------
114 * Removed the vptr of modular integer and univariate polynomial
115 classes in order to fix some crashes in that sector of CLN. Code
116 using this needs to be recompiled.
118 * Many more operator signatures make the integration of CLN types with
124 * Several compilation fixes.
126 * Made it possible to cross-compile CLN.
129 2004-11-03, version 1.1.9
130 =========================
135 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
136 complexity as opposed to O(N^2). Useful for all kinds of persistency.
138 Implementation changes
139 ----------------------
141 * Fixed several bugs in the integer input and output routines that could be
142 blamed for random crashes in the following cases: output in base 32 for
143 quite large numbers, input in base 2 for fixnums and input in base 3 for
144 fixnums on a 64 bit target.
146 * Fixed crash when radix specifiers were used in input streams.
148 * Speed up on x86_64 and ia64 by adding some inline assembly.
153 * Fixes for compilation on MacOS X and little endian Mips.
156 2004-07-01, version 1.1.8
157 =========================
159 Implementation changes
160 ----------------------
162 * Fix bug in log(a,b) when b is 1/n.
164 * Fix crash in shared library on AMD64.
169 * Fixes for some language conformance issues.
172 2004-05-02, version 1.1.7
173 =========================
175 Implementation changes
176 ----------------------
178 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
180 * Workaround GCC 3.[123] compiler bug.
182 * Parallel build should work reliably now.
185 2004-01-01, version 1.1.6
186 =========================
192 `sintL ldegree(const cl_UP& x)'.
193 (Suggested by Munagala Ramanath.)
195 Implementation changes
196 ----------------------
198 * Fixed a bug in polynomial subtraction.
199 (Reported by Munagala Ramanath.)
204 * Provide a package control file cln.pc.
206 * Updated infrastructure to autoconf-2.57.
208 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
211 2002-05-28, version 1.1.5
212 =========================
217 * Fix compilation issues with GCC-3.1.
220 2002-01-04, version 1.1.4
221 =========================
226 * Fix compilation issues with GMP-4.0 and on IBM s390.
228 * Updated infrastructure to autoconf-2.52.
231 2001-11-05, version 1.1.3
232 =========================
234 Implementation changes
235 ----------------------
237 * Fix a computation error of sin and sinh for small arguments.
238 (Reported by Christopher Kennedy.)
240 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
242 * -DNO_ASM works properly on Sparc.
245 2001-07-25, version 1.1.2
247 Implementation changes
248 ----------------------
250 * Minor cleanups and portability issues.
251 (No need to upgrade if you have been happy so far.)
254 2001-05-31, version 1.1.1
255 =========================
257 Implementation changes
258 ----------------------
260 * Minor cleanups for GCC 3.0 and libstdc++-v3.
262 * Fixes several little bugs and portability issues.
265 2000-12-14, version 1.1.0
266 =========================
271 * ISO/IEC 14882 fashion adjustments:
272 Put everything into namespace cln. All fundamental data types still
273 carry their old names. Other non-macro identifiers are now written as
274 cln::foo instead of cl_foo, except where the cl_ comes from a data type
275 (as in cl_I_to_int()). Headers are installed into a separate
276 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
277 must be manually ported to the new style. We apologize for the
280 Implementation changes
281 ----------------------
283 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
284 Configuring with --with-gmp now relies on an installed libgmp library
285 version 3 or above. We recommend libgmp-3.1 or newer.
287 * Adjusted several break-even points to make better use of a faster libgmp and
288 better match present-day CPUs.
290 * Fix several errors for architectures with 64-bit wide words.
291 (Others than Alpha, which worked already.)
293 * Fix compilation errors with current GCC-3.0 snapshots.
298 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
302 2000-01-13, version 1.0.3
303 =========================
309 `cl_I doublefactorial (uintL n)'.
311 Implementation changes
312 ----------------------
314 * Fix several little configuration errors.
316 * Fix some compilation errors with gcc-2.96 prereleases.
319 1999-09-07, version 1.0.2
320 =========================
326 `cl_I numerator (const cl_RA&)'
327 `cl_I denominator (const cl_RA&)'.
328 Suggested by Richard Kreckel and Sylvain Pion.
330 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
331 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
332 Suggested by Richard Kreckel.
334 Implementation changes
335 ----------------------
337 * Fix an endless loop when either of the functions
338 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
339 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
340 was called with an integer argument. Thanks to Richard Kreckel.
342 * Fix a bug in the addition and subtraction of rational numbers which could
343 cause unnormalized rational numbers like 3/1 to be created.
346 1999-06-09, version 1.0.1
347 =========================
352 * Speed up the functions `logand', `lognand', `logandc2'.
354 Implementation changes
355 ----------------------
357 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
359 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
360 and SPARC with LiDIA.
362 * Fix compilation errors on Win32. Thanks to Mumit Khan.
365 1999-01-12, version 1.0
366 =======================
371 * New include files, containing I/O functions:
375 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
378 The file <cl_io.h> now contains only I/O functions for characters and
381 * To access the base ring of a univariate polynomial ring, now use
382 `R->basering()' instead of `R->basering'.
384 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
387 * Implement `abs' for the class cl_RA.
389 * Replaced `read_number' with specialized functions `read_complex',
390 `read_real', `read_float', `read_rational', `read_integer'.
392 * Replaced the functions
393 `void fprint (cl_ostream stream, int x)'
394 `void fprint (cl_ostream stream, unsigned int x)'
395 `void fprint (cl_ostream stream, long x)'
396 `void fprint (cl_ostream stream, unsigned long x)'
398 `void fprintdecimal (cl_ostream stream, int x)'
399 `void fprintdecimal (cl_ostream stream, unsigned int x)'
400 `void fprintdecimal (cl_ostream stream, long x)'
401 `void fprintdecimal (cl_ostream stream, unsigned long x)'
406 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
407 now normalizes the resulting u and v to be of minimum absolute value.
408 (Suggested by the LiDIA group.)
410 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
411 the resulting float depending on the number of significant digits present
414 * Speed up the functions
415 `cl_R operator* (const cl_RA&, const cl_LF&)'
416 `cl_R operator* (const cl_LF&, const cl_RA&)'
417 `cl_R operator/ (const cl_RA&, const cl_LF&)'
418 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
420 Implementation changes
421 ----------------------
423 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
425 (Reported by Thomas Papanikolaou.)
427 * Fixed a memory leak in the long-float truncate function.
428 (Reported by Keith Briggs.)
430 * Fixed a memory leak in the modular integer exponentiation.
432 * Reduced the stack space consumption. The default stack size (typically
433 8 MB on Unix) should now be sufficient in order to run all CLN programs.
435 * Portability fixes for m68k platforms.
436 (Reported and fixed by Roman Hodek.)
441 * Restructured the sources. The subpackages
444 - base + integer + rational
445 - base + integer + rational + float
446 - base + integer + rational + float + real
447 - base + integer + rational + float + real + complex
448 - base + integer + modinteger
449 are now self-contained.
451 * The configuration script can be called with option `--enable-shared',
452 to build CLN as a shared library.
454 * Updated the LiDIA interface.
457 1998-05-07, version 0.98
458 ========================
460 Implementation changes
461 ----------------------
463 * Portability fixes for 64-bit platforms like DEC Alpha.
464 (Reported by John Cremona and Thomas Papanikolaou.)
473 * Portability fixes for new ISO/ANSI C++ compilers.
475 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
482 Implementation changes
483 ----------------------
485 * Portability fix for i686 platforms.
486 (Reported by Michael Pfeifer.)
496 <cl_ring.h>, declaring general rings,
497 <cl_GV.h>, <cl_SV.h>, defining vectors,
498 <cl_univpoly.h>, defining univariate polynomial rings.
503 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
505 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
507 Implementation changes
508 ----------------------
510 * Portability fix for 64-bit platforms like DEC Alpha:
511 Fixed a bug: endless loop in `sqrt' for large precision.
512 (Reported by Blake Jones.)
514 * Portability fix for RS/6000 platforms.
523 * Speed up the computation of the elementary transcendental functions:
524 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
525 log (for 100 digits) by about 25%,
526 sin, cos (for 1000 or 10000 digits) by about 20%,
527 asin, acos, atan (for 1000 digits) by about 25%.
529 Implementation changes
530 ----------------------
532 * Portability fix for SunOS 4 and other BSD platforms.
541 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
543 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
544 for moderate precision (< 11370 digits) were correct.
545 (Reported by Thomas Papanikolaou.)
547 Implementation changes
548 ----------------------
550 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
551 (g++ 2.8.0 or newer on HPPA).
553 * New method `debug_print()', for pretty printing of CLN objects, intended
554 to be called from the debugger.
563 * New functions `float cl_float_approx (const cl_R&)',
564 `double cl_double_approx (const cl_R&)'
565 for converting a real number to a C `float' or `double'.
570 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
572 Implementation changes
573 ----------------------
575 * New function `cl_print', for pretty printing of CLN objects, intended to be
576 called from the debugger.
578 * Portability fixes for 64-bit platforms like DEC Alpha.
588 <cl_string.h>, defining strings,
589 <cl_symbol.h>, defining symbols,
590 <cl_proplist.h>, defining property lists,
591 <cl_condition.h>, defining conditions/exceptions.
593 * New functions `cl_F cl_catalanconst ()',
594 `cl_F cl_catalanconst (const cl_F&)',
595 `cl_F cl_catalanconst (cl_float_format_t)'
596 which return Catalan's constant.
597 (Suggested by Thomas Papanikolaou.)
599 * New functions `cl_F cl_zeta (int s)',
600 `cl_F cl_zeta (int s, const cl_F&)',
601 `cl_F cl_zeta (int s, cl_float_format_t)'
602 which return the Riemann zeta function at an integral point s>1.
604 * New functions `cl_F cl_exp1 ()',
605 `cl_F cl_exp1 (const cl_F&)',
606 `cl_F cl_exp1 (cl_float_format_t)'
607 which return e = exp(1).
609 * New function `cl_I binomial (uintL n, uintL k)' which returns the
610 binomial coefficient (n choose k).
612 * New functions `int cl_I_to_int (const cl_I&)',
613 `unsigned int cl_I_to_uint (const cl_I&)',
614 `long cl_I_to_long (const cl_I&)',
615 `unsigned long cl_I_to_ulong (const cl_I&)'
616 for converting an integer to a C `int' or `long'.
618 * New functions `float cl_float_approx (const cl_I&)',
619 `float cl_float_approx (const cl_RA&)',
620 `double cl_double_approx (const cl_I&)',
621 `double cl_double_approx (const cl_RA&)'
622 for converting a rational number to a C `float' or `double'.
624 Implementation changes
625 ----------------------
627 * Moved the sources to subdirectories.
628 (Suggested by Jörg Arndt.)
633 * New benchmark for better comparing LiDIA, Pari and CLN.
635 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
636 (Suggested by Thomas Papanikolaou.)
638 * Added an ILOG Talk interface, for interactive use of the library.
647 * New include file <cl_modinteger.h>, defining modular integer rings.
649 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
650 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
655 * Speed up square root of large numbers by use of Newton's algorithm.
657 * Speed up multiplication and division of large long-floats by small integers.
659 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
660 asin, acos, atan, asinh, acosh, atanh.
661 All of them are now computed with asymptotically fast algorithms, of
662 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
664 * Fixed several bugs in the transcendental functions routines. In particular,
665 the `log' function went into an endless loop.
667 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
668 correctly for negative integers.
677 * New include file <cl_numtheory.h>, containing number theoretic functions,
678 for now only the Jacobi symbol.
683 * Speed up squaring of large numbers by 30%.
685 * Speed up division of large numbers by use of Newton's algorithm. The
686 improvement is noticeable for numbers with at least about 1000 words =
687 32000 bits = 10000 decimal digits.
689 * Speed up the binary-to-decimal conversion of large numbers by use of
690 a divide-and-conquer method. The improvement is noticeable for numbers
691 of at least 1250 words = 40000 bits = 12000 decimal digits.
693 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
694 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
696 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
697 didn't work correctly if both arguments were bignums.
699 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
700 as `mod(-n*m,m)' (n>0), returned m instead of 0.
702 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
703 values for arguments x = 2^n, n >= 29.
705 Implementation changes
706 ----------------------
708 * Speed up CLN by using the low-level addition/subtraction, shift and
709 multiplication loops from GNU gmp-2.0.2.
711 * Fixed a severe reference counting bug in the constructors
712 `type::type (const long)' and `type::type (const unsigned long)'
713 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
716 * CLN's word sequences can be chosen to be stored big-endian in memory
717 (like it was up to now) or little-endian (needed for interfacing to GMP).
726 * New include file <cl_timing.h>, contains timing facilities.
731 * Speed up the multiplication of large numbers by use of an FFT based
732 algorithm (Schönhage-Strassen). The improvement is noticeable when both
733 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
735 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
736 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
737 of 1.5. Simple trick.
739 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
740 trick from Henri Cohen.
742 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
743 which occurred when one of the arguments is zero.
745 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
747 Implementation changes
748 ----------------------
750 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
751 (Reported by Tito Flagella.)
753 * Fixed linking error when compiling without -O.
754 (Reported by Tito Flagella.)
756 * Fixed syntax error in <cl_input.h>.
757 (Reported by Tito Flagella.)
759 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
760 (Reported by Marcus Daniels.)
762 * More portability fixes.
764 * Configuration now supports the --srcdir option.
765 (Reported by Marcus Daniels.)
771 * Compiles well on Sparc, using gcc.
773 * Mail to beta testers.
779 * Compiles well on Linux, using gcc.