1 2006-08-08, version 1.1.13
2 ==========================
4 * Compilation fixes for 64-bit brokenness introduced in last release.
7 2006-08-06, version 1.1.12
8 ==========================
10 Implementation changes
11 ----------------------
13 * Fix rare assertion when printing quite large floats.
18 * Compilation fixes for several platforms: *BSD, Intel Mac, and MinGW.
21 2005-11-23, version 1.1.11
22 ==========================
27 * Considerably improved performance of number input.
30 2005-10-22, version 1.1.10
31 ==========================
33 Implementation changes
34 ----------------------
36 * Removed the vptr of modular integer and univariate polynomial
37 classes in order to fix some crashes in that sector of CLN. Code
38 using this needs to be recompiled.
40 * Many more operator signatures make the integration of CLN types with
46 * Several compilation fixes.
48 * Made it possible to cross-compile CLN.
51 2004-11-03, version 1.1.9
52 =========================
57 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
58 complexity as opposed to O(N^2). Useful for all kinds of persistency.
60 Implementation changes
61 ----------------------
63 * Fixed several bugs in the integer input and output routines that could be
64 blamed for random crashes in the following cases: output in base 32 for
65 quite large numbers, input in base 2 for fixnums and input in base 3 for
66 fixnums on a 64 bit target.
68 * Fixed crash when radix specifiers were used in input streams.
70 * Speed up on x86_64 and ia64 by adding some inline assembly.
75 * Fixes for compilation on MacOS X and little endian Mips.
78 2004-07-01, version 1.1.8
79 =========================
81 Implementation changes
82 ----------------------
84 * Fix bug in log(a,b) when b is 1/n.
86 * Fix crash in shared library on AMD64.
91 * Fixes for some language conformance issues.
94 2004-05-02, version 1.1.7
95 =========================
97 Implementation changes
98 ----------------------
100 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
102 * Workaround GCC 3.[123] compiler bug.
104 * Parallel build should work reliably now.
107 2004-01-01, version 1.1.6
108 =========================
114 `sintL ldegree(const cl_UP& x)'.
115 (Suggested by Munagala Ramanath.)
117 Implementation changes
118 ----------------------
120 * Fixed a bug in polynomial subtraction.
121 (Reported by Munagala Ramanath.)
126 * Provide a package control file cln.pc.
128 * Updated infrastructure to autoconf-2.57.
130 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
133 2002-05-28, version 1.1.5
134 =========================
139 * Fix compilation issues with GCC-3.1.
142 2002-01-04, version 1.1.4
143 =========================
148 * Fix compilation issues with GMP-4.0 and on IBM s390.
150 * Updated infrastructure to autoconf-2.52.
153 2001-11-05, version 1.1.3
154 =========================
156 Implementation changes
157 ----------------------
159 * Fix a computation error of sin and sinh for small arguments.
160 (Reported by Christopher Kennedy.)
162 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
164 * -DNO_ASM works properly on Sparc.
167 2001-07-25, version 1.1.2
169 Implementation changes
170 ----------------------
172 * Minor cleanups and portability issues.
173 (No need to upgrade if you have been happy so far.)
176 2001-05-31, version 1.1.1
177 =========================
179 Implementation changes
180 ----------------------
182 * Minor cleanups for GCC 3.0 and libstdc++-v3.
184 * Fixes several little bugs and portability issues.
187 2000-12-14, version 1.1.0
188 =========================
193 * ISO/IEC 14882 fashion adjustments:
194 Put everything into namespace cln. All fundamental data types still
195 carry their old names. Other non-macro identifiers are now written as
196 cln::foo instead of cl_foo, except where the cl_ comes from a data type
197 (as in cl_I_to_int()). Headers are installed into a separate
198 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
199 must be manually ported to the new style. We apologize for the
202 Implementation changes
203 ----------------------
205 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
206 Configuring with --with-gmp now relies on an installed libgmp library
207 version 3 or above. We recommend libgmp-3.1 or newer.
209 * Adjusted several break-even points to make better use of a faster libgmp and
210 better match present-day CPUs.
212 * Fix several errors for architectures with 64-bit wide words.
213 (Others than Alpha, which worked already.)
215 * Fix compilation errors with current GCC-3.0 snapshots.
220 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
224 2000-01-13, version 1.0.3
225 =========================
231 `cl_I doublefactorial (uintL n)'.
233 Implementation changes
234 ----------------------
236 * Fix several little configuration errors.
238 * Fix some compilation errors with gcc-2.96 prereleases.
241 1999-09-07, version 1.0.2
242 =========================
248 `cl_I numerator (const cl_RA&)'
249 `cl_I denominator (const cl_RA&)'.
250 Suggested by Richard Kreckel and Sylvain Pion.
252 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
253 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
254 Suggested by Richard Kreckel.
256 Implementation changes
257 ----------------------
259 * Fix an endless loop when either of the functions
260 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
261 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
262 was called with an integer argument. Thanks to Richard Kreckel.
264 * Fix a bug in the addition and subtraction of rational numbers which could
265 cause unnormalized rational numbers like 3/1 to be created.
268 1999-06-09, version 1.0.1
269 =========================
274 * Speed up the functions `logand', `lognand', `logandc2'.
276 Implementation changes
277 ----------------------
279 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
281 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
282 and SPARC with LiDIA.
284 * Fix compilation errors on Win32. Thanks to Mumit Khan.
287 1999-01-12, version 1.0
288 =======================
293 * New include files, containing I/O functions:
297 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
300 The file <cl_io.h> now contains only I/O functions for characters and
303 * To access the base ring of a univariate polynomial ring, now use
304 `R->basering()' instead of `R->basering'.
306 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
309 * Implement `abs' for the class cl_RA.
311 * Replaced `read_number' with specialized functions `read_complex',
312 `read_real', `read_float', `read_rational', `read_integer'.
314 * Replaced the functions
315 `void fprint (cl_ostream stream, int x)'
316 `void fprint (cl_ostream stream, unsigned int x)'
317 `void fprint (cl_ostream stream, long x)'
318 `void fprint (cl_ostream stream, unsigned long x)'
320 `void fprintdecimal (cl_ostream stream, int x)'
321 `void fprintdecimal (cl_ostream stream, unsigned int x)'
322 `void fprintdecimal (cl_ostream stream, long x)'
323 `void fprintdecimal (cl_ostream stream, unsigned long x)'
328 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
329 now normalizes the resulting u and v to be of minimum absolute value.
330 (Suggested by the LiDIA group.)
332 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
333 the resulting float depending on the number of significant digits present
336 * Speed up the functions
337 `cl_R operator* (const cl_RA&, const cl_LF&)'
338 `cl_R operator* (const cl_LF&, const cl_RA&)'
339 `cl_R operator/ (const cl_RA&, const cl_LF&)'
340 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
342 Implementation changes
343 ----------------------
345 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
347 (Reported by Thomas Papanikolaou.)
349 * Fixed a memory leak in the long-float truncate function.
350 (Reported by Keith Briggs.)
352 * Fixed a memory leak in the modular integer exponentiation.
354 * Reduced the stack space consumption. The default stack size (typically
355 8 MB on Unix) should now be sufficient in order to run all CLN programs.
357 * Portability fixes for m68k platforms.
358 (Reported and fixed by Roman Hodek.)
363 * Restructured the sources. The subpackages
366 - base + integer + rational
367 - base + integer + rational + float
368 - base + integer + rational + float + real
369 - base + integer + rational + float + real + complex
370 - base + integer + modinteger
371 are now self-contained.
373 * The configuration script can be called with option `--enable-shared',
374 to build CLN as a shared library.
376 * Updated the LiDIA interface.
379 1998-05-07, version 0.98
380 ========================
382 Implementation changes
383 ----------------------
385 * Portability fixes for 64-bit platforms like DEC Alpha.
386 (Reported by John Cremona and Thomas Papanikolaou.)
395 * Portability fixes for new ISO/ANSI C++ compilers.
397 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
404 Implementation changes
405 ----------------------
407 * Portability fix for i686 platforms.
408 (Reported by Michael Pfeifer.)
418 <cl_ring.h>, declaring general rings,
419 <cl_GV.h>, <cl_SV.h>, defining vectors,
420 <cl_univpoly.h>, defining univariate polynomial rings.
425 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
427 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
429 Implementation changes
430 ----------------------
432 * Portability fix for 64-bit platforms like DEC Alpha:
433 Fixed a bug: endless loop in `sqrt' for large precision.
434 (Reported by Blake Jones.)
436 * Portability fix for RS/6000 platforms.
445 * Speed up the computation of the elementary transcendental functions:
446 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
447 log (for 100 digits) by about 25%,
448 sin, cos (for 1000 or 10000 digits) by about 20%,
449 asin, acos, atan (for 1000 digits) by about 25%.
451 Implementation changes
452 ----------------------
454 * Portability fix for SunOS 4 and other BSD platforms.
463 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
465 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
466 for moderate precision (< 11370 digits) were correct.
467 (Reported by Thomas Papanikolaou.)
469 Implementation changes
470 ----------------------
472 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
473 (g++ 2.8.0 or newer on HPPA).
475 * New method `debug_print()', for pretty printing of CLN objects, intended
476 to be called from the debugger.
485 * New functions `float cl_float_approx (const cl_R&)',
486 `double cl_double_approx (const cl_R&)'
487 for converting a real number to a C `float' or `double'.
492 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
494 Implementation changes
495 ----------------------
497 * New function `cl_print', for pretty printing of CLN objects, intended to be
498 called from the debugger.
500 * Portability fixes for 64-bit platforms like DEC Alpha.
510 <cl_string.h>, defining strings,
511 <cl_symbol.h>, defining symbols,
512 <cl_proplist.h>, defining property lists,
513 <cl_condition.h>, defining conditions/exceptions.
515 * New functions `cl_F cl_catalanconst ()',
516 `cl_F cl_catalanconst (const cl_F&)',
517 `cl_F cl_catalanconst (cl_float_format_t)'
518 which return Catalan's constant.
519 (Suggested by Thomas Papanikolaou.)
521 * New functions `cl_F cl_zeta (int s)',
522 `cl_F cl_zeta (int s, const cl_F&)',
523 `cl_F cl_zeta (int s, cl_float_format_t)'
524 which return the Riemann zeta function at an integral point s>1.
526 * New functions `cl_F cl_exp1 ()',
527 `cl_F cl_exp1 (const cl_F&)',
528 `cl_F cl_exp1 (cl_float_format_t)'
529 which return e = exp(1).
531 * New function `cl_I binomial (uintL n, uintL k)' which returns the
532 binomial coefficient (n choose k).
534 * New functions `int cl_I_to_int (const cl_I&)',
535 `unsigned int cl_I_to_uint (const cl_I&)',
536 `long cl_I_to_long (const cl_I&)',
537 `unsigned long cl_I_to_ulong (const cl_I&)'
538 for converting an integer to a C `int' or `long'.
540 * New functions `float cl_float_approx (const cl_I&)',
541 `float cl_float_approx (const cl_RA&)',
542 `double cl_double_approx (const cl_I&)',
543 `double cl_double_approx (const cl_RA&)'
544 for converting a rational number to a C `float' or `double'.
546 Implementation changes
547 ----------------------
549 * Moved the sources to subdirectories.
550 (Suggested by Jörg Arndt.)
555 * New benchmark for better comparing LiDIA, Pari and CLN.
557 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
558 (Suggested by Thomas Papanikolaou.)
560 * Added an ILOG Talk interface, for interactive use of the library.
569 * New include file <cl_modinteger.h>, defining modular integer rings.
571 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
572 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
577 * Speed up square root of large numbers by use of Newton's algorithm.
579 * Speed up multiplication and division of large long-floats by small integers.
581 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
582 asin, acos, atan, asinh, acosh, atanh.
583 All of them are now computed with asymptotically fast algorithms, of
584 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
586 * Fixed several bugs in the transcendental functions routines. In particular,
587 the `log' function went into an endless loop.
589 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
590 correctly for negative integers.
599 * New include file <cl_numtheory.h>, containing number theoretic functions,
600 for now only the Jacobi symbol.
605 * Speed up squaring of large numbers by 30%.
607 * Speed up division of large numbers by use of Newton's algorithm. The
608 improvement is noticeable for numbers with at least about 1000 words =
609 32000 bits = 10000 decimal digits.
611 * Speed up the binary-to-decimal conversion of large numbers by use of
612 a divide-and-conquer method. The improvement is noticeable for numbers
613 of at least 1250 words = 40000 bits = 12000 decimal digits.
615 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
616 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
618 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
619 didn't work correctly if both arguments were bignums.
621 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
622 as `mod(-n*m,m)' (n>0), returned m instead of 0.
624 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
625 values for arguments x = 2^n, n >= 29.
627 Implementation changes
628 ----------------------
630 * Speed up CLN by using the low-level addition/subtraction, shift and
631 multiplication loops from GNU gmp-2.0.2.
633 * Fixed a severe reference counting bug in the constructors
634 `type::type (const long)' and `type::type (const unsigned long)'
635 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
638 * CLN's word sequences can be chosen to be stored big-endian in memory
639 (like it was up to now) or little-endian (needed for interfacing to GMP).
648 * New include file <cl_timing.h>, contains timing facilities.
653 * Speed up the multiplication of large numbers by use of an FFT based
654 algorithm (Schönhage-Strassen). The improvement is noticeable when both
655 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
657 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
658 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
659 of 1.5. Simple trick.
661 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
662 trick from Henri Cohen.
664 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
665 which occurred when one of the arguments is zero.
667 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
669 Implementation changes
670 ----------------------
672 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
673 (Reported by Tito Flagella.)
675 * Fixed linking error when compiling without -O.
676 (Reported by Tito Flagella.)
678 * Fixed syntax error in <cl_input.h>.
679 (Reported by Tito Flagella.)
681 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
682 (Reported by Marcus Daniels.)
684 * More portability fixes.
686 * Configuration now supports the --srcdir option.
687 (Reported by Marcus Daniels.)
693 * Compiles well on Sparc, using gcc.
695 * Mail to beta testers.
701 * Compiles well on Linux, using gcc.