1 2008-04-05, version 1.2.2
2 =========================
7 * Re-establish CLN-1.2.0 ABI and fix ARM build, both inadvertently broken in
11 2008-03-24, version 1.2.1
12 =========================
14 Implementation changes
15 ----------------------
17 * Fixed some bugs in the output of numbers larger than 2^32 decimal digits.
22 * Modifying C/C++ operators like +=, ++, etc. are now enabled by default.
25 2008-01-19, version 1.2.0
26 =========================
31 * Save big amounts of memory in computation of some functions and constants by:
32 - Avoiding pre-computation of series terms and instead computing them in a
34 - Avoiding computation with excess precision in binary splitting algorithm
35 by coercion of intermediate integer values to floating-point as early as
38 Implementation changes
39 ----------------------
41 * Added support for huge numbers:
42 - intC used for all counter loops.
43 - intE is now a 64-bit type on all 64-bit platforms and even on selected
46 * CLN now uses C++ exceptions for error handling. The cl_abort() hook is not
47 supported any more. Please refer to the documentation to learn about existing
50 * Fixed a bug on i386 where comparing Z/2Z ring zeros returnd random results.
52 * Removed cl_boolean. Use built-in C++ bool instead.
57 * Dropped the cln-config script. Please use pkg-config instead.
59 * Updated infrastructure to that of libtool-1.5.24.
61 * Changed encoding of all files to UTF-8.
63 * Fix compilation issues with GCC-4.3.
65 * Fix linking issues on platforms that don't feature weak symbols (like win32).
68 2006-08-08, version 1.1.13
69 ==========================
71 * Compilation fixes for 64-bit brokenness introduced in last release.
74 2006-08-06, version 1.1.12
75 ==========================
77 Implementation changes
78 ----------------------
80 * Fix rare assertion when printing quite large floats.
85 * Compilation fixes for several platforms: *BSD, Intel Mac, and MinGW.
88 2005-11-23, version 1.1.11
89 ==========================
94 * Considerably improved performance of number input.
97 2005-10-22, version 1.1.10
98 ==========================
100 Implementation changes
101 ----------------------
103 * Removed the vptr of modular integer and univariate polynomial
104 classes in order to fix some crashes in that sector of CLN. Code
105 using this needs to be recompiled.
107 * Many more operator signatures make the integration of CLN types with
113 * Several compilation fixes.
115 * Made it possible to cross-compile CLN.
118 2004-11-03, version 1.1.9
119 =========================
124 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
125 complexity as opposed to O(N^2). Useful for all kinds of persistency.
127 Implementation changes
128 ----------------------
130 * Fixed several bugs in the integer input and output routines that could be
131 blamed for random crashes in the following cases: output in base 32 for
132 quite large numbers, input in base 2 for fixnums and input in base 3 for
133 fixnums on a 64 bit target.
135 * Fixed crash when radix specifiers were used in input streams.
137 * Speed up on x86_64 and ia64 by adding some inline assembly.
142 * Fixes for compilation on MacOS X and little endian Mips.
145 2004-07-01, version 1.1.8
146 =========================
148 Implementation changes
149 ----------------------
151 * Fix bug in log(a,b) when b is 1/n.
153 * Fix crash in shared library on AMD64.
158 * Fixes for some language conformance issues.
161 2004-05-02, version 1.1.7
162 =========================
164 Implementation changes
165 ----------------------
167 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
169 * Workaround GCC 3.[123] compiler bug.
171 * Parallel build should work reliably now.
174 2004-01-01, version 1.1.6
175 =========================
181 `sintL ldegree(const cl_UP& x)'.
182 (Suggested by Munagala Ramanath.)
184 Implementation changes
185 ----------------------
187 * Fixed a bug in polynomial subtraction.
188 (Reported by Munagala Ramanath.)
193 * Provide a package control file cln.pc.
195 * Updated infrastructure to autoconf-2.57.
197 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
200 2002-05-28, version 1.1.5
201 =========================
206 * Fix compilation issues with GCC-3.1.
209 2002-01-04, version 1.1.4
210 =========================
215 * Fix compilation issues with GMP-4.0 and on IBM s390.
217 * Updated infrastructure to autoconf-2.52.
220 2001-11-05, version 1.1.3
221 =========================
223 Implementation changes
224 ----------------------
226 * Fix a computation error of sin and sinh for small arguments.
227 (Reported by Christopher Kennedy.)
229 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
231 * -DNO_ASM works properly on Sparc.
234 2001-07-25, version 1.1.2
236 Implementation changes
237 ----------------------
239 * Minor cleanups and portability issues.
240 (No need to upgrade if you have been happy so far.)
243 2001-05-31, version 1.1.1
244 =========================
246 Implementation changes
247 ----------------------
249 * Minor cleanups for GCC 3.0 and libstdc++-v3.
251 * Fixes several little bugs and portability issues.
254 2000-12-14, version 1.1.0
255 =========================
260 * ISO/IEC 14882 fashion adjustments:
261 Put everything into namespace cln. All fundamental data types still
262 carry their old names. Other non-macro identifiers are now written as
263 cln::foo instead of cl_foo, except where the cl_ comes from a data type
264 (as in cl_I_to_int()). Headers are installed into a separate
265 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
266 must be manually ported to the new style. We apologize for the
269 Implementation changes
270 ----------------------
272 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
273 Configuring with --with-gmp now relies on an installed libgmp library
274 version 3 or above. We recommend libgmp-3.1 or newer.
276 * Adjusted several break-even points to make better use of a faster libgmp and
277 better match present-day CPUs.
279 * Fix several errors for architectures with 64-bit wide words.
280 (Others than Alpha, which worked already.)
282 * Fix compilation errors with current GCC-3.0 snapshots.
287 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
291 2000-01-13, version 1.0.3
292 =========================
298 `cl_I doublefactorial (uintL n)'.
300 Implementation changes
301 ----------------------
303 * Fix several little configuration errors.
305 * Fix some compilation errors with gcc-2.96 prereleases.
308 1999-09-07, version 1.0.2
309 =========================
315 `cl_I numerator (const cl_RA&)'
316 `cl_I denominator (const cl_RA&)'.
317 Suggested by Richard Kreckel and Sylvain Pion.
319 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
320 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
321 Suggested by Richard Kreckel.
323 Implementation changes
324 ----------------------
326 * Fix an endless loop when either of the functions
327 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
328 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
329 was called with an integer argument. Thanks to Richard Kreckel.
331 * Fix a bug in the addition and subtraction of rational numbers which could
332 cause unnormalized rational numbers like 3/1 to be created.
335 1999-06-09, version 1.0.1
336 =========================
341 * Speed up the functions `logand', `lognand', `logandc2'.
343 Implementation changes
344 ----------------------
346 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
348 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
349 and SPARC with LiDIA.
351 * Fix compilation errors on Win32. Thanks to Mumit Khan.
354 1999-01-12, version 1.0
355 =======================
360 * New include files, containing I/O functions:
364 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
367 The file <cl_io.h> now contains only I/O functions for characters and
370 * To access the base ring of a univariate polynomial ring, now use
371 `R->basering()' instead of `R->basering'.
373 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
376 * Implement `abs' for the class cl_RA.
378 * Replaced `read_number' with specialized functions `read_complex',
379 `read_real', `read_float', `read_rational', `read_integer'.
381 * Replaced the functions
382 `void fprint (cl_ostream stream, int x)'
383 `void fprint (cl_ostream stream, unsigned int x)'
384 `void fprint (cl_ostream stream, long x)'
385 `void fprint (cl_ostream stream, unsigned long x)'
387 `void fprintdecimal (cl_ostream stream, int x)'
388 `void fprintdecimal (cl_ostream stream, unsigned int x)'
389 `void fprintdecimal (cl_ostream stream, long x)'
390 `void fprintdecimal (cl_ostream stream, unsigned long x)'
395 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
396 now normalizes the resulting u and v to be of minimum absolute value.
397 (Suggested by the LiDIA group.)
399 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
400 the resulting float depending on the number of significant digits present
403 * Speed up the functions
404 `cl_R operator* (const cl_RA&, const cl_LF&)'
405 `cl_R operator* (const cl_LF&, const cl_RA&)'
406 `cl_R operator/ (const cl_RA&, const cl_LF&)'
407 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
409 Implementation changes
410 ----------------------
412 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
414 (Reported by Thomas Papanikolaou.)
416 * Fixed a memory leak in the long-float truncate function.
417 (Reported by Keith Briggs.)
419 * Fixed a memory leak in the modular integer exponentiation.
421 * Reduced the stack space consumption. The default stack size (typically
422 8 MB on Unix) should now be sufficient in order to run all CLN programs.
424 * Portability fixes for m68k platforms.
425 (Reported and fixed by Roman Hodek.)
430 * Restructured the sources. The subpackages
433 - base + integer + rational
434 - base + integer + rational + float
435 - base + integer + rational + float + real
436 - base + integer + rational + float + real + complex
437 - base + integer + modinteger
438 are now self-contained.
440 * The configuration script can be called with option `--enable-shared',
441 to build CLN as a shared library.
443 * Updated the LiDIA interface.
446 1998-05-07, version 0.98
447 ========================
449 Implementation changes
450 ----------------------
452 * Portability fixes for 64-bit platforms like DEC Alpha.
453 (Reported by John Cremona and Thomas Papanikolaou.)
462 * Portability fixes for new ISO/ANSI C++ compilers.
464 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
471 Implementation changes
472 ----------------------
474 * Portability fix for i686 platforms.
475 (Reported by Michael Pfeifer.)
485 <cl_ring.h>, declaring general rings,
486 <cl_GV.h>, <cl_SV.h>, defining vectors,
487 <cl_univpoly.h>, defining univariate polynomial rings.
492 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
494 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
496 Implementation changes
497 ----------------------
499 * Portability fix for 64-bit platforms like DEC Alpha:
500 Fixed a bug: endless loop in `sqrt' for large precision.
501 (Reported by Blake Jones.)
503 * Portability fix for RS/6000 platforms.
512 * Speed up the computation of the elementary transcendental functions:
513 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
514 log (for 100 digits) by about 25%,
515 sin, cos (for 1000 or 10000 digits) by about 20%,
516 asin, acos, atan (for 1000 digits) by about 25%.
518 Implementation changes
519 ----------------------
521 * Portability fix for SunOS 4 and other BSD platforms.
530 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
532 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
533 for moderate precision (< 11370 digits) were correct.
534 (Reported by Thomas Papanikolaou.)
536 Implementation changes
537 ----------------------
539 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
540 (g++ 2.8.0 or newer on HPPA).
542 * New method `debug_print()', for pretty printing of CLN objects, intended
543 to be called from the debugger.
552 * New functions `float cl_float_approx (const cl_R&)',
553 `double cl_double_approx (const cl_R&)'
554 for converting a real number to a C `float' or `double'.
559 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
561 Implementation changes
562 ----------------------
564 * New function `cl_print', for pretty printing of CLN objects, intended to be
565 called from the debugger.
567 * Portability fixes for 64-bit platforms like DEC Alpha.
577 <cl_string.h>, defining strings,
578 <cl_symbol.h>, defining symbols,
579 <cl_proplist.h>, defining property lists,
580 <cl_condition.h>, defining conditions/exceptions.
582 * New functions `cl_F cl_catalanconst ()',
583 `cl_F cl_catalanconst (const cl_F&)',
584 `cl_F cl_catalanconst (cl_float_format_t)'
585 which return Catalan's constant.
586 (Suggested by Thomas Papanikolaou.)
588 * New functions `cl_F cl_zeta (int s)',
589 `cl_F cl_zeta (int s, const cl_F&)',
590 `cl_F cl_zeta (int s, cl_float_format_t)'
591 which return the Riemann zeta function at an integral point s>1.
593 * New functions `cl_F cl_exp1 ()',
594 `cl_F cl_exp1 (const cl_F&)',
595 `cl_F cl_exp1 (cl_float_format_t)'
596 which return e = exp(1).
598 * New function `cl_I binomial (uintL n, uintL k)' which returns the
599 binomial coefficient (n choose k).
601 * New functions `int cl_I_to_int (const cl_I&)',
602 `unsigned int cl_I_to_uint (const cl_I&)',
603 `long cl_I_to_long (const cl_I&)',
604 `unsigned long cl_I_to_ulong (const cl_I&)'
605 for converting an integer to a C `int' or `long'.
607 * New functions `float cl_float_approx (const cl_I&)',
608 `float cl_float_approx (const cl_RA&)',
609 `double cl_double_approx (const cl_I&)',
610 `double cl_double_approx (const cl_RA&)'
611 for converting a rational number to a C `float' or `double'.
613 Implementation changes
614 ----------------------
616 * Moved the sources to subdirectories.
617 (Suggested by Jörg Arndt.)
622 * New benchmark for better comparing LiDIA, Pari and CLN.
624 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
625 (Suggested by Thomas Papanikolaou.)
627 * Added an ILOG Talk interface, for interactive use of the library.
636 * New include file <cl_modinteger.h>, defining modular integer rings.
638 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
639 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
644 * Speed up square root of large numbers by use of Newton's algorithm.
646 * Speed up multiplication and division of large long-floats by small integers.
648 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
649 asin, acos, atan, asinh, acosh, atanh.
650 All of them are now computed with asymptotically fast algorithms, of
651 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
653 * Fixed several bugs in the transcendental functions routines. In particular,
654 the `log' function went into an endless loop.
656 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
657 correctly for negative integers.
666 * New include file <cl_numtheory.h>, containing number theoretic functions,
667 for now only the Jacobi symbol.
672 * Speed up squaring of large numbers by 30%.
674 * Speed up division of large numbers by use of Newton's algorithm. The
675 improvement is noticeable for numbers with at least about 1000 words =
676 32000 bits = 10000 decimal digits.
678 * Speed up the binary-to-decimal conversion of large numbers by use of
679 a divide-and-conquer method. The improvement is noticeable for numbers
680 of at least 1250 words = 40000 bits = 12000 decimal digits.
682 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
683 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
685 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
686 didn't work correctly if both arguments were bignums.
688 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
689 as `mod(-n*m,m)' (n>0), returned m instead of 0.
691 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
692 values for arguments x = 2^n, n >= 29.
694 Implementation changes
695 ----------------------
697 * Speed up CLN by using the low-level addition/subtraction, shift and
698 multiplication loops from GNU gmp-2.0.2.
700 * Fixed a severe reference counting bug in the constructors
701 `type::type (const long)' and `type::type (const unsigned long)'
702 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
705 * CLN's word sequences can be chosen to be stored big-endian in memory
706 (like it was up to now) or little-endian (needed for interfacing to GMP).
715 * New include file <cl_timing.h>, contains timing facilities.
720 * Speed up the multiplication of large numbers by use of an FFT based
721 algorithm (Schönhage-Strassen). The improvement is noticeable when both
722 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
724 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
725 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
726 of 1.5. Simple trick.
728 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
729 trick from Henri Cohen.
731 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
732 which occurred when one of the arguments is zero.
734 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
736 Implementation changes
737 ----------------------
739 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
740 (Reported by Tito Flagella.)
742 * Fixed linking error when compiling without -O.
743 (Reported by Tito Flagella.)
745 * Fixed syntax error in <cl_input.h>.
746 (Reported by Tito Flagella.)
748 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
749 (Reported by Marcus Daniels.)
751 * More portability fixes.
753 * Configuration now supports the --srcdir option.
754 (Reported by Marcus Daniels.)
760 * Compiles well on Sparc, using gcc.
762 * Mail to beta testers.
768 * Compiles well on Linux, using gcc.