1 2009-xx-xx, version 1.3.0
2 =========================
7 * Implemented a more portable module dependency mechanism.
10 2008-04-05, version 1.2.2
11 =========================
13 Implementation changes
14 ----------------------
16 * Re-establish CLN-1.2.0 ABI and fix ARM build, both inadvertently broken in
20 2008-03-24, version 1.2.1
21 =========================
23 Implementation changes
24 ----------------------
26 * Fixed some bugs in the output of numbers larger than 2^32 decimal digits.
31 * Modifying C/C++ operators like +=, ++, etc. are now enabled by default.
34 2008-01-19, version 1.2.0
35 =========================
40 * Save big amounts of memory in computation of some functions and constants by:
41 - Avoiding pre-computation of series terms and instead computing them in a
43 - Avoiding computation with excess precision in binary splitting algorithm
44 by coercion of intermediate integer values to floating-point as early as
47 Implementation changes
48 ----------------------
50 * Added support for huge numbers:
51 - intC used for all counter loops.
52 - intE is now a 64-bit type on all 64-bit platforms and even on selected
55 * CLN now uses C++ exceptions for error handling. The cl_abort() hook is not
56 supported any more. Please refer to the documentation to learn about existing
59 * Fixed a bug on i386 where comparing Z/2Z ring zeros returnd random results.
61 * Removed cl_boolean. Use built-in C++ bool instead.
66 * Dropped the cln-config script. Please use pkg-config instead.
68 * Updated infrastructure to that of libtool-1.5.24.
70 * Changed encoding of all files to UTF-8.
72 * Fix compilation issues with GCC-4.3.
74 * Fix linking issues on platforms that don't feature weak symbols (like win32).
77 2006-08-08, version 1.1.13
78 ==========================
80 * Compilation fixes for 64-bit brokenness introduced in last release.
83 2006-08-06, version 1.1.12
84 ==========================
86 Implementation changes
87 ----------------------
89 * Fix rare assertion when printing quite large floats.
94 * Compilation fixes for several platforms: *BSD, Intel Mac, and MinGW.
97 2005-11-23, version 1.1.11
98 ==========================
103 * Considerably improved performance of number input.
106 2005-10-22, version 1.1.10
107 ==========================
109 Implementation changes
110 ----------------------
112 * Removed the vptr of modular integer and univariate polynomial
113 classes in order to fix some crashes in that sector of CLN. Code
114 using this needs to be recompiled.
116 * Many more operator signatures make the integration of CLN types with
122 * Several compilation fixes.
124 * Made it possible to cross-compile CLN.
127 2004-11-03, version 1.1.9
128 =========================
133 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
134 complexity as opposed to O(N^2). Useful for all kinds of persistency.
136 Implementation changes
137 ----------------------
139 * Fixed several bugs in the integer input and output routines that could be
140 blamed for random crashes in the following cases: output in base 32 for
141 quite large numbers, input in base 2 for fixnums and input in base 3 for
142 fixnums on a 64 bit target.
144 * Fixed crash when radix specifiers were used in input streams.
146 * Speed up on x86_64 and ia64 by adding some inline assembly.
151 * Fixes for compilation on MacOS X and little endian Mips.
154 2004-07-01, version 1.1.8
155 =========================
157 Implementation changes
158 ----------------------
160 * Fix bug in log(a,b) when b is 1/n.
162 * Fix crash in shared library on AMD64.
167 * Fixes for some language conformance issues.
170 2004-05-02, version 1.1.7
171 =========================
173 Implementation changes
174 ----------------------
176 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
178 * Workaround GCC 3.[123] compiler bug.
180 * Parallel build should work reliably now.
183 2004-01-01, version 1.1.6
184 =========================
190 `sintL ldegree(const cl_UP& x)'.
191 (Suggested by Munagala Ramanath.)
193 Implementation changes
194 ----------------------
196 * Fixed a bug in polynomial subtraction.
197 (Reported by Munagala Ramanath.)
202 * Provide a package control file cln.pc.
204 * Updated infrastructure to autoconf-2.57.
206 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
209 2002-05-28, version 1.1.5
210 =========================
215 * Fix compilation issues with GCC-3.1.
218 2002-01-04, version 1.1.4
219 =========================
224 * Fix compilation issues with GMP-4.0 and on IBM s390.
226 * Updated infrastructure to autoconf-2.52.
229 2001-11-05, version 1.1.3
230 =========================
232 Implementation changes
233 ----------------------
235 * Fix a computation error of sin and sinh for small arguments.
236 (Reported by Christopher Kennedy.)
238 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
240 * -DNO_ASM works properly on Sparc.
243 2001-07-25, version 1.1.2
245 Implementation changes
246 ----------------------
248 * Minor cleanups and portability issues.
249 (No need to upgrade if you have been happy so far.)
252 2001-05-31, version 1.1.1
253 =========================
255 Implementation changes
256 ----------------------
258 * Minor cleanups for GCC 3.0 and libstdc++-v3.
260 * Fixes several little bugs and portability issues.
263 2000-12-14, version 1.1.0
264 =========================
269 * ISO/IEC 14882 fashion adjustments:
270 Put everything into namespace cln. All fundamental data types still
271 carry their old names. Other non-macro identifiers are now written as
272 cln::foo instead of cl_foo, except where the cl_ comes from a data type
273 (as in cl_I_to_int()). Headers are installed into a separate
274 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
275 must be manually ported to the new style. We apologize for the
278 Implementation changes
279 ----------------------
281 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
282 Configuring with --with-gmp now relies on an installed libgmp library
283 version 3 or above. We recommend libgmp-3.1 or newer.
285 * Adjusted several break-even points to make better use of a faster libgmp and
286 better match present-day CPUs.
288 * Fix several errors for architectures with 64-bit wide words.
289 (Others than Alpha, which worked already.)
291 * Fix compilation errors with current GCC-3.0 snapshots.
296 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
300 2000-01-13, version 1.0.3
301 =========================
307 `cl_I doublefactorial (uintL n)'.
309 Implementation changes
310 ----------------------
312 * Fix several little configuration errors.
314 * Fix some compilation errors with gcc-2.96 prereleases.
317 1999-09-07, version 1.0.2
318 =========================
324 `cl_I numerator (const cl_RA&)'
325 `cl_I denominator (const cl_RA&)'.
326 Suggested by Richard Kreckel and Sylvain Pion.
328 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
329 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
330 Suggested by Richard Kreckel.
332 Implementation changes
333 ----------------------
335 * Fix an endless loop when either of the functions
336 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
337 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
338 was called with an integer argument. Thanks to Richard Kreckel.
340 * Fix a bug in the addition and subtraction of rational numbers which could
341 cause unnormalized rational numbers like 3/1 to be created.
344 1999-06-09, version 1.0.1
345 =========================
350 * Speed up the functions `logand', `lognand', `logandc2'.
352 Implementation changes
353 ----------------------
355 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
357 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
358 and SPARC with LiDIA.
360 * Fix compilation errors on Win32. Thanks to Mumit Khan.
363 1999-01-12, version 1.0
364 =======================
369 * New include files, containing I/O functions:
373 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
376 The file <cl_io.h> now contains only I/O functions for characters and
379 * To access the base ring of a univariate polynomial ring, now use
380 `R->basering()' instead of `R->basering'.
382 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
385 * Implement `abs' for the class cl_RA.
387 * Replaced `read_number' with specialized functions `read_complex',
388 `read_real', `read_float', `read_rational', `read_integer'.
390 * Replaced the functions
391 `void fprint (cl_ostream stream, int x)'
392 `void fprint (cl_ostream stream, unsigned int x)'
393 `void fprint (cl_ostream stream, long x)'
394 `void fprint (cl_ostream stream, unsigned long x)'
396 `void fprintdecimal (cl_ostream stream, int x)'
397 `void fprintdecimal (cl_ostream stream, unsigned int x)'
398 `void fprintdecimal (cl_ostream stream, long x)'
399 `void fprintdecimal (cl_ostream stream, unsigned long x)'
404 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
405 now normalizes the resulting u and v to be of minimum absolute value.
406 (Suggested by the LiDIA group.)
408 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
409 the resulting float depending on the number of significant digits present
412 * Speed up the functions
413 `cl_R operator* (const cl_RA&, const cl_LF&)'
414 `cl_R operator* (const cl_LF&, const cl_RA&)'
415 `cl_R operator/ (const cl_RA&, const cl_LF&)'
416 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
418 Implementation changes
419 ----------------------
421 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
423 (Reported by Thomas Papanikolaou.)
425 * Fixed a memory leak in the long-float truncate function.
426 (Reported by Keith Briggs.)
428 * Fixed a memory leak in the modular integer exponentiation.
430 * Reduced the stack space consumption. The default stack size (typically
431 8 MB on Unix) should now be sufficient in order to run all CLN programs.
433 * Portability fixes for m68k platforms.
434 (Reported and fixed by Roman Hodek.)
439 * Restructured the sources. The subpackages
442 - base + integer + rational
443 - base + integer + rational + float
444 - base + integer + rational + float + real
445 - base + integer + rational + float + real + complex
446 - base + integer + modinteger
447 are now self-contained.
449 * The configuration script can be called with option `--enable-shared',
450 to build CLN as a shared library.
452 * Updated the LiDIA interface.
455 1998-05-07, version 0.98
456 ========================
458 Implementation changes
459 ----------------------
461 * Portability fixes for 64-bit platforms like DEC Alpha.
462 (Reported by John Cremona and Thomas Papanikolaou.)
471 * Portability fixes for new ISO/ANSI C++ compilers.
473 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
480 Implementation changes
481 ----------------------
483 * Portability fix for i686 platforms.
484 (Reported by Michael Pfeifer.)
494 <cl_ring.h>, declaring general rings,
495 <cl_GV.h>, <cl_SV.h>, defining vectors,
496 <cl_univpoly.h>, defining univariate polynomial rings.
501 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
503 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
505 Implementation changes
506 ----------------------
508 * Portability fix for 64-bit platforms like DEC Alpha:
509 Fixed a bug: endless loop in `sqrt' for large precision.
510 (Reported by Blake Jones.)
512 * Portability fix for RS/6000 platforms.
521 * Speed up the computation of the elementary transcendental functions:
522 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
523 log (for 100 digits) by about 25%,
524 sin, cos (for 1000 or 10000 digits) by about 20%,
525 asin, acos, atan (for 1000 digits) by about 25%.
527 Implementation changes
528 ----------------------
530 * Portability fix for SunOS 4 and other BSD platforms.
539 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
541 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
542 for moderate precision (< 11370 digits) were correct.
543 (Reported by Thomas Papanikolaou.)
545 Implementation changes
546 ----------------------
548 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
549 (g++ 2.8.0 or newer on HPPA).
551 * New method `debug_print()', for pretty printing of CLN objects, intended
552 to be called from the debugger.
561 * New functions `float cl_float_approx (const cl_R&)',
562 `double cl_double_approx (const cl_R&)'
563 for converting a real number to a C `float' or `double'.
568 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
570 Implementation changes
571 ----------------------
573 * New function `cl_print', for pretty printing of CLN objects, intended to be
574 called from the debugger.
576 * Portability fixes for 64-bit platforms like DEC Alpha.
586 <cl_string.h>, defining strings,
587 <cl_symbol.h>, defining symbols,
588 <cl_proplist.h>, defining property lists,
589 <cl_condition.h>, defining conditions/exceptions.
591 * New functions `cl_F cl_catalanconst ()',
592 `cl_F cl_catalanconst (const cl_F&)',
593 `cl_F cl_catalanconst (cl_float_format_t)'
594 which return Catalan's constant.
595 (Suggested by Thomas Papanikolaou.)
597 * New functions `cl_F cl_zeta (int s)',
598 `cl_F cl_zeta (int s, const cl_F&)',
599 `cl_F cl_zeta (int s, cl_float_format_t)'
600 which return the Riemann zeta function at an integral point s>1.
602 * New functions `cl_F cl_exp1 ()',
603 `cl_F cl_exp1 (const cl_F&)',
604 `cl_F cl_exp1 (cl_float_format_t)'
605 which return e = exp(1).
607 * New function `cl_I binomial (uintL n, uintL k)' which returns the
608 binomial coefficient (n choose k).
610 * New functions `int cl_I_to_int (const cl_I&)',
611 `unsigned int cl_I_to_uint (const cl_I&)',
612 `long cl_I_to_long (const cl_I&)',
613 `unsigned long cl_I_to_ulong (const cl_I&)'
614 for converting an integer to a C `int' or `long'.
616 * New functions `float cl_float_approx (const cl_I&)',
617 `float cl_float_approx (const cl_RA&)',
618 `double cl_double_approx (const cl_I&)',
619 `double cl_double_approx (const cl_RA&)'
620 for converting a rational number to a C `float' or `double'.
622 Implementation changes
623 ----------------------
625 * Moved the sources to subdirectories.
626 (Suggested by Jörg Arndt.)
631 * New benchmark for better comparing LiDIA, Pari and CLN.
633 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
634 (Suggested by Thomas Papanikolaou.)
636 * Added an ILOG Talk interface, for interactive use of the library.
645 * New include file <cl_modinteger.h>, defining modular integer rings.
647 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
648 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
653 * Speed up square root of large numbers by use of Newton's algorithm.
655 * Speed up multiplication and division of large long-floats by small integers.
657 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
658 asin, acos, atan, asinh, acosh, atanh.
659 All of them are now computed with asymptotically fast algorithms, of
660 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
662 * Fixed several bugs in the transcendental functions routines. In particular,
663 the `log' function went into an endless loop.
665 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
666 correctly for negative integers.
675 * New include file <cl_numtheory.h>, containing number theoretic functions,
676 for now only the Jacobi symbol.
681 * Speed up squaring of large numbers by 30%.
683 * Speed up division of large numbers by use of Newton's algorithm. The
684 improvement is noticeable for numbers with at least about 1000 words =
685 32000 bits = 10000 decimal digits.
687 * Speed up the binary-to-decimal conversion of large numbers by use of
688 a divide-and-conquer method. The improvement is noticeable for numbers
689 of at least 1250 words = 40000 bits = 12000 decimal digits.
691 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
692 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
694 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
695 didn't work correctly if both arguments were bignums.
697 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
698 as `mod(-n*m,m)' (n>0), returned m instead of 0.
700 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
701 values for arguments x = 2^n, n >= 29.
703 Implementation changes
704 ----------------------
706 * Speed up CLN by using the low-level addition/subtraction, shift and
707 multiplication loops from GNU gmp-2.0.2.
709 * Fixed a severe reference counting bug in the constructors
710 `type::type (const long)' and `type::type (const unsigned long)'
711 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
714 * CLN's word sequences can be chosen to be stored big-endian in memory
715 (like it was up to now) or little-endian (needed for interfacing to GMP).
724 * New include file <cl_timing.h>, contains timing facilities.
729 * Speed up the multiplication of large numbers by use of an FFT based
730 algorithm (Schönhage-Strassen). The improvement is noticeable when both
731 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
733 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
734 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
735 of 1.5. Simple trick.
737 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
738 trick from Henri Cohen.
740 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
741 which occurred when one of the arguments is zero.
743 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
745 Implementation changes
746 ----------------------
748 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
749 (Reported by Tito Flagella.)
751 * Fixed linking error when compiling without -O.
752 (Reported by Tito Flagella.)
754 * Fixed syntax error in <cl_input.h>.
755 (Reported by Tito Flagella.)
757 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
758 (Reported by Marcus Daniels.)
760 * More portability fixes.
762 * Configuration now supports the --srcdir option.
763 (Reported by Marcus Daniels.)
769 * Compiles well on Sparc, using gcc.
771 * Mail to beta testers.
777 * Compiles well on Linux, using gcc.