1 2005-11-23, version 1.1.11
2 ==========================
7 * Considerably improved performance of number input.
10 2005-10-22, version 1.1.10
11 ==========================
13 Implementation changes
14 ----------------------
16 * Removed the vptr of modular integer and univariate polynomial
17 classes in order to fix some crashes in that sector of CLN. Code
18 using this needs to be recompiled.
20 * Many more operator signatures make the integration of CLN types with
26 * Several compilation fixes.
28 * Made it possible to cross-compile CLN.
31 2004-11-03, version 1.1.9
32 =========================
37 * Input of numbers in bases 2, 4, 8, 16 and 32 is now done in linear bit
38 complexity as opposed to O(N^2). Useful for all kinds of persistency.
40 Implementation changes
41 ----------------------
43 * Fixed several bugs in the integer input and output routines that could be
44 blamed for random crashes in the following cases: output in base 32 for
45 quite large numbers, input in base 2 for fixnums and input in base 3 for
46 fixnums on a 64 bit target.
48 * Fixed crash when radix specifiers were used in input streams.
50 * Speed up on x86_64 and ia64 by adding some inline assembly.
55 * Fixes for compilation on MacOS X and little endian Mips.
58 2004-07-01, version 1.1.8
59 =========================
61 Implementation changes
62 ----------------------
64 * Fix bug in log(a,b) when b is 1/n.
66 * Fix crash in shared library on AMD64.
71 * Fixes for some language conformance issues.
74 2004-05-02, version 1.1.7
75 =========================
77 Implementation changes
78 ----------------------
80 * Fixed two corner case bugs in cl_LF multiplication and sqrt.
82 * Workaround GCC 3.[123] compiler bug.
84 * Parallel build should work reliably now.
87 2004-01-01, version 1.1.6
88 =========================
94 `sintL ldegree(const cl_UP& x)'.
95 (Suggested by Munagala Ramanath.)
97 Implementation changes
98 ----------------------
100 * Fixed a bug in polynomial subtraction.
101 (Reported by Munagala Ramanath.)
106 * Provide a package control file cln.pc.
108 * Updated infrastructure to autoconf-2.57.
110 * Fixed compilation issues with GCC-3.4 prereleases and on Mac OSX.
113 2002-05-28, version 1.1.5
114 =========================
119 * Fix compilation issues with GCC-3.1.
122 2002-01-04, version 1.1.4
123 =========================
128 * Fix compilation issues with GMP-4.0 and on IBM s390.
130 * Updated infrastructure to autoconf-2.52.
133 2001-11-05, version 1.1.3
134 =========================
136 Implementation changes
137 ----------------------
139 * Fix a computation error of sin and sinh for small arguments.
140 (Reported by Christopher Kennedy.)
142 * Module ordering works with gcc-3.0 -fuse-cxa-atexit now.
144 * -DNO_ASM works properly on Sparc.
147 2001-07-25, version 1.1.2
149 Implementation changes
150 ----------------------
152 * Minor cleanups and portability issues.
153 (No need to upgrade if you have been happy so far.)
156 2001-05-31, version 1.1.1
157 =========================
159 Implementation changes
160 ----------------------
162 * Minor cleanups for GCC 3.0 and libstdc++-v3.
164 * Fixes several little bugs and portability issues.
167 2000-12-14, version 1.1.0
168 =========================
173 * ISO/IEC 14882 fashion adjustments:
174 Put everything into namespace cln. All fundamental data types still
175 carry their old names. Other non-macro identifiers are now written as
176 cln::foo instead of cl_foo, except where the cl_ comes from a data type
177 (as in cl_I_to_int()). Headers are installed into a separate
178 directory, #include <cln/foo.h> instead of <cl_foo.h>. Applications
179 must be manually ported to the new style. We apologize for the
182 Implementation changes
183 ----------------------
185 * Removed internal copy of mpn subdirectory tree from libgmp-2.0.2.
186 Configuring with --with-gmp now relies on an installed libgmp library
187 version 3 or above. We recommend libgmp-3.1 or newer.
189 * Adjusted several break-even points to make better use of a faster libgmp and
190 better match present-day CPUs.
192 * Fix several errors for architectures with 64-bit wide words.
193 (Others than Alpha, which worked already.)
195 * Fix compilation errors with current GCC-3.0 snapshots.
200 * Added package tools: script cln-config and autoconf macro AC_PATH_CLN
204 2000-01-13, version 1.0.3
205 =========================
211 `cl_I doublefactorial (uintL n)'.
213 Implementation changes
214 ----------------------
216 * Fix several little configuration errors.
218 * Fix some compilation errors with gcc-2.96 prereleases.
221 1999-09-07, version 1.0.2
222 =========================
228 `cl_I numerator (const cl_RA&)'
229 `cl_I denominator (const cl_RA&)'.
230 Suggested by Richard Kreckel and Sylvain Pion.
232 * New function `cl_equal_hashcode' for the classes cl_N, cl_R, cl_RA, cl_I,
233 cl_F, cl_SF, cl_FF, cl_DF, cl_LF.
234 Suggested by Richard Kreckel.
236 Implementation changes
237 ----------------------
239 * Fix an endless loop when either of the functions
240 `cl_boolean rootp (const cl_RA&, uintL, cl_RA*)',
241 `cl_boolean rootp (const cl_RA&, const cl_I&, cl_RA*)'
242 was called with an integer argument. Thanks to Richard Kreckel.
244 * Fix a bug in the addition and subtraction of rational numbers which could
245 cause unnormalized rational numbers like 3/1 to be created.
248 1999-06-09, version 1.0.1
249 =========================
254 * Speed up the functions `logand', `lognand', `logandc2'.
256 Implementation changes
257 ----------------------
259 * Fix compilation errors with gcc-2.7.2, egcs-1.1.2 and gcc-2.95.
261 * Fix compilation errors on HPPA, MIPS, some versions of DEC Alpha, OpenBSD,
262 and SPARC with LiDIA.
264 * Fix compilation errors on Win32. Thanks to Mumit Khan.
267 1999-01-12, version 1.0
268 =======================
273 * New include files, containing I/O functions:
277 <cl_sfloat_io.h> <cl_ffloat_io.h> <cl_dfloat_io.h> <cl_lfloat_io.h>
280 The file <cl_io.h> now contains only I/O functions for characters and
283 * To access the base ring of a univariate polynomial ring, now use
284 `R->basering()' instead of `R->basering'.
286 * Implement `plusp', `max', `min' for the classes cl_F, cl_SF, cl_FF, cl_DF,
289 * Implement `abs' for the class cl_RA.
291 * Replaced `read_number' with specialized functions `read_complex',
292 `read_real', `read_float', `read_rational', `read_integer'.
294 * Replaced the functions
295 `void fprint (cl_ostream stream, int x)'
296 `void fprint (cl_ostream stream, unsigned int x)'
297 `void fprint (cl_ostream stream, long x)'
298 `void fprint (cl_ostream stream, unsigned long x)'
300 `void fprintdecimal (cl_ostream stream, int x)'
301 `void fprintdecimal (cl_ostream stream, unsigned int x)'
302 `void fprintdecimal (cl_ostream stream, long x)'
303 `void fprintdecimal (cl_ostream stream, unsigned long x)'
308 * The function `cl_I xgcd (const cl_I& a, const cl_I& b, cl_I* u, cl_I* v)'
309 now normalizes the resulting u and v to be of minimum absolute value.
310 (Suggested by the LiDIA group.)
312 * The conversion from string to cl_F, cl_R, cl_N now chooses the format of
313 the resulting float depending on the number of significant digits present
316 * Speed up the functions
317 `cl_R operator* (const cl_RA&, const cl_LF&)'
318 `cl_R operator* (const cl_LF&, const cl_RA&)'
319 `cl_R operator/ (const cl_RA&, const cl_LF&)'
320 `cl_LF operator/ (const cl_LF&, const cl_RA&)'
322 Implementation changes
323 ----------------------
325 * Fix for `cl_I ash (const cl_I& x, const cl_I& n)' when |n| is between
327 (Reported by Thomas Papanikolaou.)
329 * Fixed a memory leak in the long-float truncate function.
330 (Reported by Keith Briggs.)
332 * Fixed a memory leak in the modular integer exponentiation.
334 * Reduced the stack space consumption. The default stack size (typically
335 8 MB on Unix) should now be sufficient in order to run all CLN programs.
337 * Portability fixes for m68k platforms.
338 (Reported and fixed by Roman Hodek.)
343 * Restructured the sources. The subpackages
346 - base + integer + rational
347 - base + integer + rational + float
348 - base + integer + rational + float + real
349 - base + integer + rational + float + real + complex
350 - base + integer + modinteger
351 are now self-contained.
353 * The configuration script can be called with option `--enable-shared',
354 to build CLN as a shared library.
356 * Updated the LiDIA interface.
359 1998-05-07, version 0.98
360 ========================
362 Implementation changes
363 ----------------------
365 * Portability fixes for 64-bit platforms like DEC Alpha.
366 (Reported by John Cremona and Thomas Papanikolaou.)
375 * Portability fixes for new ISO/ANSI C++ compilers.
377 * Updated the LiDIA interface for LiDIA version 1.3 and fixed a couple of
384 Implementation changes
385 ----------------------
387 * Portability fix for i686 platforms.
388 (Reported by Michael Pfeifer.)
398 <cl_ring.h>, declaring general rings,
399 <cl_GV.h>, <cl_SV.h>, defining vectors,
400 <cl_univpoly.h>, defining univariate polynomial rings.
405 * Speed up the internal computation of ln(2) and ln(10) by about 20%.
407 * Speed up the computation of exp (for 1000 digits or more) by about 10%.
409 Implementation changes
410 ----------------------
412 * Portability fix for 64-bit platforms like DEC Alpha:
413 Fixed a bug: endless loop in `sqrt' for large precision.
414 (Reported by Blake Jones.)
416 * Portability fix for RS/6000 platforms.
425 * Speed up the computation of the elementary transcendental functions:
426 exp, sinh, cosh, asinh, atanh (for 100 digits) by about 30%,
427 log (for 100 digits) by about 25%,
428 sin, cos (for 1000 or 10000 digits) by about 20%,
429 asin, acos, atan (for 1000 digits) by about 25%.
431 Implementation changes
432 ----------------------
434 * Portability fix for SunOS 4 and other BSD platforms.
443 * Fixed a bug: endless recursion in `cl_boolean sqrtp (const cl_RA&, cl_RA*)'.
445 * Fixed a bug: Only the first 85% of the digits returned by `cl_eulerconst'
446 for moderate precision (< 11370 digits) were correct.
447 (Reported by Thomas Papanikolaou.)
449 Implementation changes
450 ----------------------
452 * Fix static initialization order bug. Requires g++ 2.7.0 or newer
453 (g++ 2.8.0 or newer on HPPA).
455 * New method `debug_print()', for pretty printing of CLN objects, intended
456 to be called from the debugger.
465 * New functions `float cl_float_approx (const cl_R&)',
466 `double cl_double_approx (const cl_R&)'
467 for converting a real number to a C `float' or `double'.
472 * Speed up `cl_zeta' using Cohen-Villegas-Zagier convergence acceleration.
474 Implementation changes
475 ----------------------
477 * New function `cl_print', for pretty printing of CLN objects, intended to be
478 called from the debugger.
480 * Portability fixes for 64-bit platforms like DEC Alpha.
490 <cl_string.h>, defining strings,
491 <cl_symbol.h>, defining symbols,
492 <cl_proplist.h>, defining property lists,
493 <cl_condition.h>, defining conditions/exceptions.
495 * New functions `cl_F cl_catalanconst ()',
496 `cl_F cl_catalanconst (const cl_F&)',
497 `cl_F cl_catalanconst (cl_float_format_t)'
498 which return Catalan's constant.
499 (Suggested by Thomas Papanikolaou.)
501 * New functions `cl_F cl_zeta (int s)',
502 `cl_F cl_zeta (int s, const cl_F&)',
503 `cl_F cl_zeta (int s, cl_float_format_t)'
504 which return the Riemann zeta function at an integral point s>1.
506 * New functions `cl_F cl_exp1 ()',
507 `cl_F cl_exp1 (const cl_F&)',
508 `cl_F cl_exp1 (cl_float_format_t)'
509 which return e = exp(1).
511 * New function `cl_I binomial (uintL n, uintL k)' which returns the
512 binomial coefficient (n choose k).
514 * New functions `int cl_I_to_int (const cl_I&)',
515 `unsigned int cl_I_to_uint (const cl_I&)',
516 `long cl_I_to_long (const cl_I&)',
517 `unsigned long cl_I_to_ulong (const cl_I&)'
518 for converting an integer to a C `int' or `long'.
520 * New functions `float cl_float_approx (const cl_I&)',
521 `float cl_float_approx (const cl_RA&)',
522 `double cl_double_approx (const cl_I&)',
523 `double cl_double_approx (const cl_RA&)'
524 for converting a rational number to a C `float' or `double'.
526 Implementation changes
527 ----------------------
529 * Moved the sources to subdirectories.
530 (Suggested by Jörg Arndt.)
535 * New benchmark for better comparing LiDIA, Pari and CLN.
537 * Added an interface to LiDIA, allows using CLN as a replacement of libI.
538 (Suggested by Thomas Papanikolaou.)
540 * Added an ILOG Talk interface, for interactive use of the library.
549 * New include file <cl_modinteger.h>, defining modular integer rings.
551 * New functions `cl_F cl_eulerconst ()', `cl_F cl_eulerconst (const cl_F&)',
552 `cl_F cl_eulerconst (cl_float_format_t)' which return Euler's constant.
557 * Speed up square root of large numbers by use of Newton's algorithm.
559 * Speed up multiplication and division of large long-floats by small integers.
561 * Speed up the computation of pi, e, exp, log, sin, cos, tan, sinh, cosh, tanh,
562 asin, acos, atan, asinh, acosh, atanh.
563 All of them are now computed with asymptotically fast algorithms, of
564 bit complexity O(log(N)^2*M(N)) = O(N^(1+epsilon)) for N digits.
566 * Fixed several bugs in the transcendental functions routines. In particular,
567 the `log' function went into an endless loop.
569 * Fixed a bug: The cl_I -> cl_DF and cl_I -> cl_LF conversions didn't work
570 correctly for negative integers.
579 * New include file <cl_numtheory.h>, containing number theoretic functions,
580 for now only the Jacobi symbol.
585 * Speed up squaring of large numbers by 30%.
587 * Speed up division of large numbers by use of Newton's algorithm. The
588 improvement is noticeable for numbers with at least about 1000 words =
589 32000 bits = 10000 decimal digits.
591 * Speed up the binary-to-decimal conversion of large numbers by use of
592 a divide-and-conquer method. The improvement is noticeable for numbers
593 of at least 1250 words = 40000 bits = 12000 decimal digits.
595 * Fixed a bug: The functions `cl_boolean minusp (const type&)' and
596 `cl_boolean zerop (const type&)' were missing for type=cl_I and type=cl_RA.
598 * Fixed a bug: The function `cl_boolean logtest (const cl_I&, const cl_I&)'
599 didn't work correctly if both arguments were bignums.
601 * Fixed a bug: The function `cl_I mod (const cl_I&, const cl_I&)', when called
602 as `mod(-n*m,m)' (n>0), returned m instead of 0.
604 * Fixed a bug: The function `uintL power2p (const cl_I&)' returned wrong
605 values for arguments x = 2^n, n >= 29.
607 Implementation changes
608 ----------------------
610 * Speed up CLN by using the low-level addition/subtraction, shift and
611 multiplication loops from GNU gmp-2.0.2.
613 * Fixed a severe reference counting bug in the constructors
614 `type::type (const long)' and `type::type (const unsigned long)'
615 for type=cl_number, type=cl_N, type=cl_RA, type=cl_I that most often
618 * CLN's word sequences can be chosen to be stored big-endian in memory
619 (like it was up to now) or little-endian (needed for interfacing to GMP).
628 * New include file <cl_timing.h>, contains timing facilities.
633 * Speed up the multiplication of large numbers by use of an FFT based
634 algorithm (Schönhage-Strassen). The improvement is noticeable when both
635 factors have at least 1250 words = 40000 bits = 12000 decimal digits.
637 * Speed up the functions `cl_I gcd (const cl_I&, const cl_I&)' and
638 `cl_I xgcd (const cl_I&, const cl_I&, cl_I*, cl_I*)' by a factor
639 of 1.5. Simple trick.
641 * Speed up the function `cl_boolean sqrtp (const cl_I&, cl_I*)' using a
642 trick from Henri Cohen.
644 * Fixed an endless loop in the function `uint32 gcd (uint32, uint32)'
645 which occurred when one of the arguments is zero.
647 * Fixed an endless loop in the function `cl_I factorial (uintL)'.
649 Implementation changes
650 ----------------------
652 * <cl_io.h> now uses iostreams by default instead of stdio FILE pointers.
653 (Reported by Tito Flagella.)
655 * Fixed linking error when compiling without -O.
656 (Reported by Tito Flagella.)
658 * Fixed syntax error in <cl_input.h>.
659 (Reported by Tito Flagella.)
661 * Fixed syntax errors in src/cl_DS.h and src/cl_FF_plus.h.
662 (Reported by Marcus Daniels.)
664 * More portability fixes.
666 * Configuration now supports the --srcdir option.
667 (Reported by Marcus Daniels.)
673 * Compiles well on Sparc, using gcc.
675 * Mail to beta testers.
681 * Compiles well on Linux, using gcc.