1 2004-11-03, version 1.1.9
2 =========================
7 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
8 complexity as opposed to O(N^2). Useful for all kinds of persistency.
10 Implementation changes
11 ----------------------
13 * Fixed several bugs in the integer input and output routines that could be
14 blamed for random crashes in the following cases: output in base 32 for
15 quite large numbers, input in base 2 for fixnums and input in base 3 for
16 fixnums on a 64 bit target.
18 * Fixed crash when radix specifiers were used in input streams.
20 * Speed up on x86_64 and ia64 by adding some inline assembly.
25 * Fixes for compilation on MacOS X and little endian Mips.
28 2004-07-01, version 1.1.8
29 =========================
31 Implementation changes
32 ----------------------
34 * Fix bug in log(a,b) when b is 1/n.
36 * Fix crash in shared library on AMD64.
41 * Fixes for some language conformance issues.
44 2004-05-02, version 1.1.7
45 =========================
47 Implementation changes
48 ----------------------
50 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
52 * Workaround GCC 3.[123] compiler bug.
54 * Parallel build should work reliably now.
57 2004-01-01, version 1.1.6
58 =========================
64 `sintL ldegree(const cl_UP& x)'.
65 (Suggested by Munagala Ramanath.)
67 Implementation changes
68 ----------------------
70 * Fixed a bug in polynomial subtraction.
71 (Reported by Munagala Ramanath.)
76 * Provide a package control file cln.pc.
78 * Updated infrastructure to autoconf-2.57.
80 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
83 2002-05-28, version 1.1.5
84 =========================
89 * Fix compilation issues with GCC-3.1.
92 2002-01-04, version 1.1.4
93 =========================
98 * Fix compilation issues with GMP-4.0 and on IBM s390.
100 * Updated infrastructure to autoconf-2.52.
103 2001-11-05, version 1.1.3
104 =========================
106 Implementation changes
107 ----------------------
109 * Fix a computation error of sin and sinh for small arguments.
110 (Reported by Christopher Kennedy.)
112 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
114 * -DNO_ASM works properly on Sparc.
117 2001-07-25, version 1.1.2
119 Implementation changes
120 ----------------------
122 * Minor cleanups and portability issues.
123 (No need to upgrade if you have been happy so far.)
126 2001-05-31, version 1.1.1
127 =========================
129 Implementation changes
130 ----------------------
132 * Minor cleanups for GCC 3.0 and libstdc++-v3.
134 * Fixes several little bugs and portability issues.
137 2000-12-14, version 1.1.0
138 =========================
143 * ISO/IEC 14882 fashion adjustments:
144 Put everything into namespace cln. All fundamental data types still
145 carry their old names. Other non-macro identifiers are now written as
146 cln::foo instead of cl_foo, except where the cl_ comes from a data type
147 (as in cl_I_to_int()). Headers are installed into a separate
148 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
149 must be manually ported to the new style. We apologize for the
152 Implementation changes
153 ----------------------
155 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
156 Configuring with --with-gmp now relies on an installed libgmp library
157 version 3 or above. We recommend libgmp-3.1 or newer.
159 * Adjusted several break-even points to make better use of a faster libgmp and
160 better match present-day CPUs.
162 * Fix several errors for architectures with 64-bit wide words.
163 (Others than Alpha, which worked already.)
165 * Fix compilation errors with current GCC-3.0 snapshots.
170 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
174 2000-01-13, version 1.0.3
175 =========================
181 `cl_I doublefactorial (uintL n)'.
183 Implementation changes
184 ----------------------
186 * Fix several little configuration errors.
188 * Fix some compilation errors with gcc-2.96 prereleases.
191 1999-09-07, version 1.0.2
192 =========================
198 `cl_I numerator (const cl_RA&)'
199 `cl_I denominator (const cl_RA&)'.
200 Suggested by Richard Kreckel and Sylvain Pion.
202 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
203 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
204 Suggested by Richard Kreckel.
206 Implementation changes
207 ----------------------
209 * Fix an endless loop when either of the functions
210 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
211 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
212 was called with an integer argument. Thanks to Richard Kreckel.
214 * Fix a bug in the addition and subtraction of rational numbers which could
215 cause unnormalized rational numbers like 3/1 to be created.
218 1999-06-09, version 1.0.1
219 =========================
224 * Speed up the functions `logand', `lognand', `logandc2'.
226 Implementation changes
227 ----------------------
229 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
231 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
232 and SPARC with LiDIA.
234 * Fix compilation errors on Win32. Thanks to Mumit Khan.
237 1999-01-12, version 1.0
238 =======================
243 * New include files, containing I/O functions:
247 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
250 The file <cl_io.h> now contains only I/O functions for characters and
253 * To access the base ring of a univariate polynomial ring, now use
254 `R->basering()' instead of `R->basering'.
256 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
259 * Implement `abs' for the class cl_RA.
261 * Replaced `read_number' with specialized functions `read_complex',
262 `read_real', `read_float', `read_rational', `read_integer'.
264 * Replaced the functions
265 `void fprint (cl_ostream stream, int x)'
266 `void fprint (cl_ostream stream, unsigned int x)'
267 `void fprint (cl_ostream stream, long x)'
268 `void fprint (cl_ostream stream, unsigned long x)'
270 `void fprintdecimal (cl_ostream stream, int x)'
271 `void fprintdecimal (cl_ostream stream, unsigned int x)'
272 `void fprintdecimal (cl_ostream stream, long x)'
273 `void fprintdecimal (cl_ostream stream, unsigned long x)'
278 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
279 now normalizes the resulting u and v to be of minimum absolute value.
280 (Suggested by the LiDIA group.)
282 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
283 the resulting float depending on the number of significant digits present
286 * Speed up the functions
287 `cl_R operator* (const cl_RA&, const cl_LF&)'
288 `cl_R operator* (const cl_LF&, const cl_RA&)'
289 `cl_R operator/ (const cl_RA&, const cl_LF&)'
290 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
292 Implementation changes
293 ----------------------
295 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
297 (Reported by Thomas Papanikolaou.)
299 * Fixed a memory leak in the long-float truncate function.
300 (Reported by Keith Briggs.)
302 * Fixed a memory leak in the modular integer exponentiation.
304 * Reduced the stack space consumption. The default stack size (typically
305 8 MB on Unix) should now be sufficient in order to run all CLN programs.
307 * Portability fixes for m68k platforms.
308 (Reported and fixed by Roman Hodek.)
313 * Restructured the sources. The subpackages
316 - base + integer + rational
317 - base + integer + rational + float
318 - base + integer + rational + float + real
319 - base + integer + rational + float + real + complex
320 - base + integer + modinteger
321 are now self-contained.
323 * The configuration script can be called with option `--enable-shared',
324 to build CLN as a shared library.
326 * Updated the LiDIA interface.
329 1998-05-07, version 0.98
330 ========================
332 Implementation changes
333 ----------------------
335 * Portability fixes for 64-bit platforms like DEC Alpha.
336 (Reported by John Cremona and Thomas Papanikolaou.)
345 * Portability fixes for new ISO/ANSI C++ compilers.
347 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
354 Implementation changes
355 ----------------------
357 * Portability fix for i686 platforms.
358 (Reported by Michael Pfeifer.)
368 <cl_ring.h>, declaring general rings,
369 <cl_GV.h>, <cl_SV.h>, defining vectors,
370 <cl_univpoly.h>, defining univariate polynomial rings.
375 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
377 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
379 Implementation changes
380 ----------------------
382 * Portability fix for 64-bit platforms like DEC Alpha:
383 Fixed a bug: endless loop in `sqrt' for large precision.
384 (Reported by Blake Jones.)
386 * Portability fix for RS/6000 platforms.
395 * Speed up the computation of the elementary transcendental functions:
396 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
397 log (for 100 digits) by about 25%,
398 sin, cos (for 1000 or 10000 digits) by about 20%,
399 asin, acos, atan (for 1000 digits) by about 25%.
401 Implementation changes
402 ----------------------
404 * Portability fix for SunOS 4 and other BSD platforms.
413 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
415 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
416 for moderate precision (< 11370 digits) were correct.
417 (Reported by Thomas Papanikolaou.)
419 Implementation changes
420 ----------------------
422 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
423 (g++ 2.8.0 or newer on HPPA).
425 * New method `debug_print()', for pretty printing of CLN objects, intended
426 to be called from the debugger.
435 * New functions `float cl_float_approx (const cl_R&)',
436 `double cl_double_approx (const cl_R&)'
437 for converting a real number to a C `float' or `double'.
442 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
444 Implementation changes
445 ----------------------
447 * New function `cl_print', for pretty printing of CLN objects, intended to be
448 called from the debugger.
450 * Portability fixes for 64-bit platforms like DEC Alpha.
460 <cl_string.h>, defining strings,
461 <cl_symbol.h>, defining symbols,
462 <cl_proplist.h>, defining property lists,
463 <cl_condition.h>, defining conditions/exceptions.
465 * New functions `cl_F cl_catalanconst ()',
466 `cl_F cl_catalanconst (const cl_F&)',
467 `cl_F cl_catalanconst (cl_float_format_t)'
468 which return Catalan's constant.
469 (Suggested by Thomas Papanikolaou.)
471 * New functions `cl_F cl_zeta (int s)',
472 `cl_F cl_zeta (int s, const cl_F&)',
473 `cl_F cl_zeta (int s, cl_float_format_t)'
474 which return the Riemann zeta function at an integral point s>1.
476 * New functions `cl_F cl_exp1 ()',
477 `cl_F cl_exp1 (const cl_F&)',
478 `cl_F cl_exp1 (cl_float_format_t)'
479 which return e = exp(1).
481 * New function `cl_I binomial (uintL n, uintL k)' which returns the
482 binomial coefficient (n choose k).
484 * New functions `int cl_I_to_int (const cl_I&)',
485 `unsigned int cl_I_to_uint (const cl_I&)',
486 `long cl_I_to_long (const cl_I&)',
487 `unsigned long cl_I_to_ulong (const cl_I&)'
488 for converting an integer to a C `int' or `long'.
490 * New functions `float cl_float_approx (const cl_I&)',
491 `float cl_float_approx (const cl_RA&)',
492 `double cl_double_approx (const cl_I&)',
493 `double cl_double_approx (const cl_RA&)'
494 for converting a rational number to a C `float' or `double'.
496 Implementation changes
497 ----------------------
499 * Moved the sources to subdirectories.
500 (Suggested by Jörg Arndt.)
505 * New benchmark for better comparing LiDIA, Pari and CLN.
507 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
508 (Suggested by Thomas Papanikolaou.)
510 * Added an ILOG Talk interface, for interactive use of the library.
519 * New include file <cl_modinteger.h>, defining modular integer rings.
521 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
522 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
527 * Speed up square root of large numbers by use of Newton's algorithm.
529 * Speed up multiplication and division of large long-floats by small integers.
531 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
532 asin, acos, atan, asinh, acosh, atanh.
533 All of them are now computed with asymptotically fast algorithms, of
534 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
536 * Fixed several bugs in the transcendental functions routines. In particular,
537 the `log' function went into an endless loop.
539 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
540 correctly for negative integers.
549 * New include file <cl_numtheory.h>, containing number theoretic functions,
550 for now only the Jacobi symbol.
555 * Speed up squaring of large numbers by 30%.
557 * Speed up division of large numbers by use of Newton's algorithm. The
558 improvement is noticeable for numbers with at least about 1000 words =
559 32000 bits = 10000 decimal digits.
561 * Speed up the binary-to-decimal conversion of large numbers by use of
562 a divide-and-conquer method. The improvement is noticeable for numbers
563 of at least 1250 words = 40000 bits = 12000 decimal digits.
565 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
566 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
568 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
569 didn't work correctly if both arguments were bignums.
571 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
572 as `mod(-n*m,m)' (n>0), returned m instead of 0.
574 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
575 values for arguments x = 2^n, n >= 29.
577 Implementation changes
578 ----------------------
580 * Speed up CLN by using the low-level addition/subtraction, shift and
581 multiplication loops from GNU gmp-2.0.2.
583 * Fixed a severe reference counting bug in the constructors
584 `type::type (const long)' and `type::type (const unsigned long)'
585 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
588 * CLN's word sequences can be chosen to be stored big-endian in memory
589 (like it was up to now) or little-endian (needed for interfacing to GMP).
598 * New include file <cl_timing.h>, contains timing facilities.
603 * Speed up the multiplication of large numbers by use of an FFT based
604 algorithm (Schönhage-Strassen). The improvement is noticeable when both
605 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
607 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
608 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
609 of 1.5. Simple trick.
611 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
612 trick from Henri Cohen.
614 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
615 which occurred when one of the arguments is zero.
617 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
619 Implementation changes
620 ----------------------
622 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
623 (Reported by Tito Flagella.)
625 * Fixed linking error when compiling without -O.
626 (Reported by Tito Flagella.)
628 * Fixed syntax error in <cl_input.h>.
629 (Reported by Tito Flagella.)
631 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
632 (Reported by Marcus Daniels.)
634 * More portability fixes.
636 * Configuration now supports the --srcdir option.
637 (Reported by Marcus Daniels.)
643 * Compiles well on Sparc, using gcc.
645 * Mail to beta testers.
651 * Compiles well on Linux, using gcc.