1 2011-05-08, version 1.3.2
2 =========================
7 * Fixed a bug in scale_float(cl_LF, cl_I) when the scale factor exceeded 2^31
13 * Improved portability to some non-GCC compilers.
16 2009-09-24, version 1.3.1
17 =========================
19 Implementation changes
20 ----------------------
22 * Fixed a crash in double_approx(cl_RA) on 64-bit platforms.
24 * Add basic support for Renesas SH (sh4).
27 2009-06-30, version 1.3.0
28 =========================
33 * Use the GNU autotools as build system.
35 * Implemented a more portable module dependency mechanism.
38 2008-04-05, version 1.2.2
39 =========================
41 Implementation changes
42 ----------------------
44 * Re-establish CLN-1.2.0 ABI and fix ARM build, both inadvertently broken in
48 2008-03-24, version 1.2.1
49 =========================
51 Implementation changes
52 ----------------------
54 * Fixed some bugs in the output of numbers larger than 2^32 decimal digits.
59 * Modifying C/C++ operators like +=, ++, etc. are now enabled by default.
62 2008-01-19, version 1.2.0
63 =========================
68 * Save big amounts of memory in computation of some functions and constants by:
69 - Avoiding pre-computation of series terms and instead computing them in a
71 - Avoiding computation with excess precision in binary splitting algorithm
72 by coercion of intermediate integer values to floating-point as early as
75 Implementation changes
76 ----------------------
78 * Added support for huge numbers:
79 - intC used for all counter loops.
80 - intE is now a 64-bit type on all 64-bit platforms and even on selected
83 * CLN now uses C++ exceptions for error handling. The cl_abort() hook is not
84 supported any more. Please refer to the documentation to learn about existing
87 * Fixed a bug on i386 where comparing Z/2Z ring zeros returnd random results.
89 * Removed cl_boolean. Use built-in C++ bool instead.
94 * Dropped the cln-config script. Please use pkg-config instead.
96 * Updated infrastructure to that of libtool-1.5.24.
98 * Changed encoding of all files to UTF-8.
100 * Fix compilation issues with GCC-4.3.
102 * Fix linking issues on platforms that don't feature weak symbols (like win32).
105 2006-08-08, version 1.1.13
106 ==========================
108 * Compilation fixes for 64-bit brokenness introduced in last release.
111 2006-08-06, version 1.1.12
112 ==========================
114 Implementation changes
115 ----------------------
117 * Fix rare assertion when printing quite large floats.
122 * Compilation fixes for several platforms: *BSD, Intel Mac, and MinGW.
125 2005-11-23, version 1.1.11
126 ==========================
131 * Considerably improved performance of number input.
134 2005-10-22, version 1.1.10
135 ==========================
137 Implementation changes
138 ----------------------
140 * Removed the vptr of modular integer and univariate polynomial
141 classes in order to fix some crashes in that sector of CLN. Code
142 using this needs to be recompiled.
144 * Many more operator signatures make the integration of CLN types with
150 * Several compilation fixes.
152 * Made it possible to cross-compile CLN.
155 2004-11-03, version 1.1.9
156 =========================
161 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
162 complexity as opposed to O(N^2). Useful for all kinds of persistency.
164 Implementation changes
165 ----------------------
167 * Fixed several bugs in the integer input and output routines that could be
168 blamed for random crashes in the following cases: output in base 32 for
169 quite large numbers, input in base 2 for fixnums and input in base 3 for
170 fixnums on a 64 bit target.
172 * Fixed crash when radix specifiers were used in input streams.
174 * Speed up on x86_64 and ia64 by adding some inline assembly.
179 * Fixes for compilation on MacOS X and little endian Mips.
182 2004-07-01, version 1.1.8
183 =========================
185 Implementation changes
186 ----------------------
188 * Fix bug in log(a,b) when b is 1/n.
190 * Fix crash in shared library on AMD64.
195 * Fixes for some language conformance issues.
198 2004-05-02, version 1.1.7
199 =========================
201 Implementation changes
202 ----------------------
204 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
206 * Workaround GCC 3.[123] compiler bug.
208 * Parallel build should work reliably now.
211 2004-01-01, version 1.1.6
212 =========================
218 `sintL ldegree(const cl_UP& x)'.
219 (Suggested by Munagala Ramanath.)
221 Implementation changes
222 ----------------------
224 * Fixed a bug in polynomial subtraction.
225 (Reported by Munagala Ramanath.)
230 * Provide a package control file cln.pc.
232 * Updated infrastructure to autoconf-2.57.
234 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
237 2002-05-28, version 1.1.5
238 =========================
243 * Fix compilation issues with GCC-3.1.
246 2002-01-04, version 1.1.4
247 =========================
252 * Fix compilation issues with GMP-4.0 and on IBM s390.
254 * Updated infrastructure to autoconf-2.52.
257 2001-11-05, version 1.1.3
258 =========================
260 Implementation changes
261 ----------------------
263 * Fix a computation error of sin and sinh for small arguments.
264 (Reported by Christopher Kennedy.)
266 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
268 * -DNO_ASM works properly on Sparc.
271 2001-07-25, version 1.1.2
273 Implementation changes
274 ----------------------
276 * Minor cleanups and portability issues.
277 (No need to upgrade if you have been happy so far.)
280 2001-05-31, version 1.1.1
281 =========================
283 Implementation changes
284 ----------------------
286 * Minor cleanups for GCC 3.0 and libstdc++-v3.
288 * Fixes several little bugs and portability issues.
291 2000-12-14, version 1.1.0
292 =========================
297 * ISO/IEC 14882 fashion adjustments:
298 Put everything into namespace cln. All fundamental data types still
299 carry their old names. Other non-macro identifiers are now written as
300 cln::foo instead of cl_foo, except where the cl_ comes from a data type
301 (as in cl_I_to_int()). Headers are installed into a separate
302 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
303 must be manually ported to the new style. We apologize for the
306 Implementation changes
307 ----------------------
309 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
310 Configuring with --with-gmp now relies on an installed libgmp library
311 version 3 or above. We recommend libgmp-3.1 or newer.
313 * Adjusted several break-even points to make better use of a faster libgmp and
314 better match present-day CPUs.
316 * Fix several errors for architectures with 64-bit wide words.
317 (Others than Alpha, which worked already.)
319 * Fix compilation errors with current GCC-3.0 snapshots.
324 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
328 2000-01-13, version 1.0.3
329 =========================
335 `cl_I doublefactorial (uintL n)'.
337 Implementation changes
338 ----------------------
340 * Fix several little configuration errors.
342 * Fix some compilation errors with gcc-2.96 prereleases.
345 1999-09-07, version 1.0.2
346 =========================
352 `cl_I numerator (const cl_RA&)'
353 `cl_I denominator (const cl_RA&)'.
354 Suggested by Richard Kreckel and Sylvain Pion.
356 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
357 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
358 Suggested by Richard Kreckel.
360 Implementation changes
361 ----------------------
363 * Fix an endless loop when either of the functions
364 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
365 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
366 was called with an integer argument. Thanks to Richard Kreckel.
368 * Fix a bug in the addition and subtraction of rational numbers which could
369 cause unnormalized rational numbers like 3/1 to be created.
372 1999-06-09, version 1.0.1
373 =========================
378 * Speed up the functions `logand', `lognand', `logandc2'.
380 Implementation changes
381 ----------------------
383 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
385 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
386 and SPARC with LiDIA.
388 * Fix compilation errors on Win32. Thanks to Mumit Khan.
391 1999-01-12, version 1.0
392 =======================
397 * New include files, containing I/O functions:
401 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
404 The file <cl_io.h> now contains only I/O functions for characters and
407 * To access the base ring of a univariate polynomial ring, now use
408 `R->basering()' instead of `R->basering'.
410 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
413 * Implement `abs' for the class cl_RA.
415 * Replaced `read_number' with specialized functions `read_complex',
416 `read_real', `read_float', `read_rational', `read_integer'.
418 * Replaced the functions
419 `void fprint (cl_ostream stream, int x)'
420 `void fprint (cl_ostream stream, unsigned int x)'
421 `void fprint (cl_ostream stream, long x)'
422 `void fprint (cl_ostream stream, unsigned long x)'
424 `void fprintdecimal (cl_ostream stream, int x)'
425 `void fprintdecimal (cl_ostream stream, unsigned int x)'
426 `void fprintdecimal (cl_ostream stream, long x)'
427 `void fprintdecimal (cl_ostream stream, unsigned long x)'
432 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
433 now normalizes the resulting u and v to be of minimum absolute value.
434 (Suggested by the LiDIA group.)
436 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
437 the resulting float depending on the number of significant digits present
440 * Speed up the functions
441 `cl_R operator* (const cl_RA&, const cl_LF&)'
442 `cl_R operator* (const cl_LF&, const cl_RA&)'
443 `cl_R operator/ (const cl_RA&, const cl_LF&)'
444 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
446 Implementation changes
447 ----------------------
449 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
451 (Reported by Thomas Papanikolaou.)
453 * Fixed a memory leak in the long-float truncate function.
454 (Reported by Keith Briggs.)
456 * Fixed a memory leak in the modular integer exponentiation.
458 * Reduced the stack space consumption. The default stack size (typically
459 8 MB on Unix) should now be sufficient in order to run all CLN programs.
461 * Portability fixes for m68k platforms.
462 (Reported and fixed by Roman Hodek.)
467 * Restructured the sources. The subpackages
470 - base + integer + rational
471 - base + integer + rational + float
472 - base + integer + rational + float + real
473 - base + integer + rational + float + real + complex
474 - base + integer + modinteger
475 are now self-contained.
477 * The configuration script can be called with option `--enable-shared',
478 to build CLN as a shared library.
480 * Updated the LiDIA interface.
483 1998-05-07, version 0.98
484 ========================
486 Implementation changes
487 ----------------------
489 * Portability fixes for 64-bit platforms like DEC Alpha.
490 (Reported by John Cremona and Thomas Papanikolaou.)
499 * Portability fixes for new ISO/ANSI C++ compilers.
501 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
508 Implementation changes
509 ----------------------
511 * Portability fix for i686 platforms.
512 (Reported by Michael Pfeifer.)
522 <cl_ring.h>, declaring general rings,
523 <cl_GV.h>, <cl_SV.h>, defining vectors,
524 <cl_univpoly.h>, defining univariate polynomial rings.
529 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
531 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
533 Implementation changes
534 ----------------------
536 * Portability fix for 64-bit platforms like DEC Alpha:
537 Fixed a bug: endless loop in `sqrt' for large precision.
538 (Reported by Blake Jones.)
540 * Portability fix for RS/6000 platforms.
549 * Speed up the computation of the elementary transcendental functions:
550 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
551 log (for 100 digits) by about 25%,
552 sin, cos (for 1000 or 10000 digits) by about 20%,
553 asin, acos, atan (for 1000 digits) by about 25%.
555 Implementation changes
556 ----------------------
558 * Portability fix for SunOS 4 and other BSD platforms.
567 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
569 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
570 for moderate precision (< 11370 digits) were correct.
571 (Reported by Thomas Papanikolaou.)
573 Implementation changes
574 ----------------------
576 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
577 (g++ 2.8.0 or newer on HPPA).
579 * New method `debug_print()', for pretty printing of CLN objects, intended
580 to be called from the debugger.
589 * New functions `float cl_float_approx (const cl_R&)',
590 `double cl_double_approx (const cl_R&)'
591 for converting a real number to a C `float' or `double'.
596 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
598 Implementation changes
599 ----------------------
601 * New function `cl_print', for pretty printing of CLN objects, intended to be
602 called from the debugger.
604 * Portability fixes for 64-bit platforms like DEC Alpha.
614 <cl_string.h>, defining strings,
615 <cl_symbol.h>, defining symbols,
616 <cl_proplist.h>, defining property lists,
617 <cl_condition.h>, defining conditions/exceptions.
619 * New functions `cl_F cl_catalanconst ()',
620 `cl_F cl_catalanconst (const cl_F&)',
621 `cl_F cl_catalanconst (cl_float_format_t)'
622 which return Catalan's constant.
623 (Suggested by Thomas Papanikolaou.)
625 * New functions `cl_F cl_zeta (int s)',
626 `cl_F cl_zeta (int s, const cl_F&)',
627 `cl_F cl_zeta (int s, cl_float_format_t)'
628 which return the Riemann zeta function at an integral point s>1.
630 * New functions `cl_F cl_exp1 ()',
631 `cl_F cl_exp1 (const cl_F&)',
632 `cl_F cl_exp1 (cl_float_format_t)'
633 which return e = exp(1).
635 * New function `cl_I binomial (uintL n, uintL k)' which returns the
636 binomial coefficient (n choose k).
638 * New functions `int cl_I_to_int (const cl_I&)',
639 `unsigned int cl_I_to_uint (const cl_I&)',
640 `long cl_I_to_long (const cl_I&)',
641 `unsigned long cl_I_to_ulong (const cl_I&)'
642 for converting an integer to a C `int' or `long'.
644 * New functions `float cl_float_approx (const cl_I&)',
645 `float cl_float_approx (const cl_RA&)',
646 `double cl_double_approx (const cl_I&)',
647 `double cl_double_approx (const cl_RA&)'
648 for converting a rational number to a C `float' or `double'.
650 Implementation changes
651 ----------------------
653 * Moved the sources to subdirectories.
654 (Suggested by Jörg Arndt.)
659 * New benchmark for better comparing LiDIA, Pari and CLN.
661 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
662 (Suggested by Thomas Papanikolaou.)
664 * Added an ILOG Talk interface, for interactive use of the library.
673 * New include file <cl_modinteger.h>, defining modular integer rings.
675 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
676 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
681 * Speed up square root of large numbers by use of Newton's algorithm.
683 * Speed up multiplication and division of large long-floats by small integers.
685 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
686 asin, acos, atan, asinh, acosh, atanh.
687 All of them are now computed with asymptotically fast algorithms, of
688 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
690 * Fixed several bugs in the transcendental functions routines. In particular,
691 the `log' function went into an endless loop.
693 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
694 correctly for negative integers.
703 * New include file <cl_numtheory.h>, containing number theoretic functions,
704 for now only the Jacobi symbol.
709 * Speed up squaring of large numbers by 30%.
711 * Speed up division of large numbers by use of Newton's algorithm. The
712 improvement is noticeable for numbers with at least about 1000 words =
713 32000 bits = 10000 decimal digits.
715 * Speed up the binary-to-decimal conversion of large numbers by use of
716 a divide-and-conquer method. The improvement is noticeable for numbers
717 of at least 1250 words = 40000 bits = 12000 decimal digits.
719 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
720 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
722 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
723 didn't work correctly if both arguments were bignums.
725 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
726 as `mod(-n*m,m)' (n>0), returned m instead of 0.
728 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
729 values for arguments x = 2^n, n >= 29.
731 Implementation changes
732 ----------------------
734 * Speed up CLN by using the low-level addition/subtraction, shift and
735 multiplication loops from GNU gmp-2.0.2.
737 * Fixed a severe reference counting bug in the constructors
738 `type::type (const long)' and `type::type (const unsigned long)'
739 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
742 * CLN's word sequences can be chosen to be stored big-endian in memory
743 (like it was up to now) or little-endian (needed for interfacing to GMP).
752 * New include file <cl_timing.h>, contains timing facilities.
757 * Speed up the multiplication of large numbers by use of an FFT based
758 algorithm (Schönhage-Strassen). The improvement is noticeable when both
759 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
761 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
762 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
763 of 1.5. Simple trick.
765 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
766 trick from Henri Cohen.
768 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
769 which occurred when one of the arguments is zero.
771 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
773 Implementation changes
774 ----------------------
776 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
777 (Reported by Tito Flagella.)
779 * Fixed linking error when compiling without -O.
780 (Reported by Tito Flagella.)
782 * Fixed syntax error in <cl_input.h>.
783 (Reported by Tito Flagella.)
785 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
786 (Reported by Marcus Daniels.)
788 * More portability fixes.
790 * Configuration now supports the --srcdir option.
791 (Reported by Marcus Daniels.)
797 * Compiles well on Sparc, using gcc.
799 * Mail to beta testers.
805 * Compiles well on Linux, using gcc.