1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987
|
2008-04-10
fixed a buffer overflow
floatnum.c:
- a temporary buffer in float_setinteger holding the ASCII
representation of an int, was 1 byte too short.
2008-03-29
fixed a semantic error in floatconfig
remove expbase from structures in floatio
floatio, floatconvert
- refactor: remove unnecessary expbase field from structures
floatconfig
- bug fix: corrected semantic errors in BITS_IN_*EXP
that could lead to failures when converting huge numbers to
non-decimal bases
2008-03-25
let most of the IO conversion code see the exponent as an integer value
(instead of an ASCII string)
floatio
- refactor: keep exponents as integers, not strings
2008-02-04
separate the formatting of the scale from that of the significand
clarify semantics of BITS_IN_EXP
floatconfig:
- clarify semantics of BITS_IN_EXP
floatio:
- return the scale as integer and provide a separate scale formatter
2008-01-30
simplify _trigreduce
bugfix in float_shl
error reporting in float_out
floattrig:
- use float_divmod to simplify calculations
floathmath:
- corrected a recently introduced bug in float_shl
floatio, floatconvert:
- float_out now reports an Error
2008-01-17
made floatnum 'const' aware.
use of enum Error instead of #define'd error codes
bug fix in _gamma0_5, when a fall back to the general gamma routine
did not handle a negative argument correctly.
cleaner error reporting, several fixes in this area
all routines in floathmath and floatnum now take const parameters
(const floatstruct*), when a parameter is invariant in a function call.
Actually, these parameters are not guaranteed to be constant; they may,
for efficiency, assume intermediate values throughout an operation. But
they are always restored on return (note: this 'constness' is *not* safe
with respect to concurrency!).
This rework revealed a few minor bugs related to error handling, not visible
to SpeedCrunch, and one, where SpeedCrunch could freeze (Gamma(x), when x is
a very large multiple of -0.5). Fixed these.
All error values now use the enum Error from errors.h
floathmath, floatnum, floatcommon, floatgamma, floatlogic:
- const awareness
- bug fix in _gamma0_5
- error constants from an enum
floatio
- error constants from an enum
2008-01-09
bugfixes in float_frac, float_gamma
floatnum.c
- float_frac could cause memory corruption, when the fractional
part had leading zeros
floatgamma.c
- pochhammer_i reported a series of poles for x == 0 and n >= 0,
where it should be regular, actually.
testfloatnum
- added a testcase checking correct handling of tiny fractional parts
2008-01-01
introduced a few const attributes
2007-12-28
refined error reporting
floatconvert
- float_in reports all IO errors now
2007-12-09
fine tuning
fixes of minor bugs
floaterf, floatseries
- fine tuning of erf and the Taylor series
- fixed a few glitches
testfloatnum
- added a test suite for float_erf
- extended the suite of float_erfc to check more corner cases
2007-12-08
bug fix
floaterf
- _erfc could hang on tiny x. Fixed
- tuned _erfc
floatseries
- adjusted working precisions to be best for _erfc
testfloatnum
- added test suite for float_erfc
2007-12-06
bug fixes
fine tuning
floaterf
- recent changes introduced a bug in _erf for large x. Fixed
floatseries
- testing revealed: erfcsum could die if the working precision fell
below 1 before the final loop was executed. Fixed.
- fine tuned the precision parameters in erfcsum
testfloatnum
- added a test suite for erfcsum
2007-12-05
fine tuning
floatseries:
- adjusted the working precision such, that the relative error of
the asymptotic series of erfc is below 1e-99 for a 100-digit
evaluation
floaterf
- sharper evaluation of the lower bound for the asymptotic series
based on a theoretical analysis
- correct error reporting in _erfc
testfloatnum
- added a test suite for erfcasymptotic
2007-12-02
fine tuning
floatseries:
- the asymptotic series for erfc accumulated round-off
errors. increased the working precision in the loop by one.
2007-11-30
bug fix
floaterf
- fixed wrong evaluation of erf for 0.5 < x < 1
2007-11-29
erf and erfc implemented, but rigorous testing and fine tuning
are not done
floaterf
- completed functions _erf and _erfc
floatconst
- several new constants and a working space for the erfc function
2007-11-28
started implementing the erf function:
Taylor series for erf
floaterf
- new file containing erf related functions
floatconst
- new constant 2/sqrt(pi) used for the erf function
floatseries
- added Taylor series for sqrt(pi)*erf(x)/2
- added asymptotic series for erfc
floathmath
- added float_erf
- added float_erfc
2007-11-20
comments
floatnum.h
- minor style change in comments
2007-10-30
relax behaviour of float_setexponent
floatnum:
- it's convenient if you can assign any exponent to a
zero (without effect), regardless of the valid exponent
range. Reworked float_setexponent accordingly.
testfloatnum:
- adjusted testfloatnum to reflect the change
2007-08-30
fix error reporting
rename a couple of functions
more and better tests
floatnum
- float_round takes a dest parameter
- fix error reporting of float_round
- fix error reporting of float_int
- rename float_changesign -> float_neg
- rename float_clone -> float_copy
testfloatnum
- added tests for float_move
- new (extended) test routine for float_round
2007-08-29
bugfix: new tests uncovered a memory leak
more tests
floatnum
- fix memory leak in float_clone
testfloatnum
- better and more tests
2007-08-28
start testing error behaviour
more tests
floatcommon
- const modifier for parameter in setasciiz
floatnum
- make float_clone sensitive to maxdigits
2007-08-26
allow modification of the precision limit
get rid of float_setbcnum: do not expose too much of the
internals in the interface
more tests
floatnum
- update comments
- simplify float_setrange
- new functions float_setprecision and float_getprecision
- remove float_setbcnum
- rewrite float_setinteger to not use float_setbcnum
testfloatnum
- remove test cases for float_setbcnum
- rewrite test_divmod so it does not use float_setbcnum
- more tests
2007-08-23
bugfix, extend testcases
a new set of test cases uncovered a bug in float_setsignificand
floatcommon
- refer to the variable maxdigits instead of the constant MAXDIGITS
floatnum
- make maxdigits mutable
- bugfix in float_setsignificand (digit count could be off by one in
rarely encountered special cases)
testfloatnum
- more test cases
2007-08-11
extend test cases
testfloatnum
- extend tests of test_getexponent to extreme values
2007-08-10
refactor, extend test cases
floatconvert
- pack2floatnum: simplify a bit, uses variable range limits now
testfloatnum
- started refactoring testcase handling
- {long,check}{add,mul}: check reverse order of arguments automatically
- bugfix: tc_checkadd, tc_checkmul: check expression was wrong, rendering
the tests useless
- added some corner case tests to test_longarray{add,mul},
test_setnan, test_setzero, test_isnan, test_iszero
2007-08-09
sometimes it would be nice if temporary results could exceed the overflow
underflow limit. Make these limits variable.
Rename this file to ChangeLog.floatnum.
floatnum
- introduce a variable overflow/underflow limit. New functions float_setrange
and float_isvalidexp
2007-08-01
comments, refactor
floatconst, floatgamma
- simplify a bit by merging two constants
floatconfig
- changed some comments
2007-07-31
refactor, comments
floatgamma
- added some comments
- minor refactoring
2007-07-30
bugfix, extension of nCr
floathmath
- extend the range of nCr towards big integers by using lnGamma
floatseries
- bugfix: lnGamma returned NaN on extreme large arguments
(> 1e43).
2007-07-29
compatibility issues, refactor
floatcommon
- new function float_roundtoint
hmath
- make trunc and round work more like current SpeedCrunch
0.8 implementation
- enable engineering format
- check unstable evaluation near poles and zeros:
underflow to zero or set NaN
floatconvert
- refactor: use float_roundtoint
2007-07-28
bugfixes
hmath
- break conditions in loop of cbrt were often missed.
rewrote cbrt and used adapting precision and
exponent extraction.
- implement fall back behaviour in formatFixed.
formatFixed falls back to scientific format, if
it cannot display large or tiny values
2007-07-27
bugfix, refactor, comments, checking memory leaks
floatconfig
- make DECPRECISION user-definable. MATHPRECISION is the maximum
precision of the math library, the user cannot change that
- made comments reflect the new meaning
hmath
- The algorithm in HMath::cbrt could overflow.
Applied some low level improvements, that not only fixed this,
but also improved performance.
- removed a define HMATH_MAX_PREC
- let + and - underflow to zero in case of almost full cancellation
floathmath
- trigonometric functions now report FLOAT_UNSTABLE, if fed with
a large value > 10^DECPRECISION
floatnum
- refactor: was no leak, but always call float_free when float_create
is called.
floatconvert
- refactor: was no leak, but always call float_free when float_create
is called.
- checked all memory allocations, seems to be free of mem leaks
floatconst
- add a floatmath_exit routine, that releases all constants
not needed so far, but want to have a complete interface
floatseries
- fixed a memory leak in an error path (that should never be reached,
but anyway)
floattrig, floathmath, floatcommon, floatexp, floatgamma, floatipower,
floatlog, floatpower, floatlong, floatio, hmath
- checked all floatnum memory allocations, seem to be free of mem leaks
2007-07-25
bugfixes, comments
refactor: finally got the ioparam stuff out of the core. It
is free of grammar information now, except that digit
sequences are ASCII sequences in big-endian format.
This is not going to be eliminated, because almost all
clients (scanners) will have digit sequences in this
format (or can easily convert it to it).
floattrig
- fixed a comparison in _arctan: An optimization was
completely skipped. Now arctan runs a bit faster when x > 1
and does not choke on an endless loop any more
when arguments are really huge (> 1e100000000)
- create an early out path in _arcsinlt1 so that this function
does not underflow on extremely tiny arguments
floathmath
- fixed comparisons in float_arcsin, float_arccos and
float_arccosxplus1. They did not recognize the error case
|x| > 1 before
floatio
- added or modified comments
- refactor: _str2seq: rename match into leadingdigit
- refactor: _int2str: remove unnecessary code, leave n unmodified,
get rid of getioparams
- refactor: _clearprefix has only one caller and is trivial, so get
rid of it
- refactor: _exp2desc: _int2str does not query ioparams any more
so there is no need to set the default ioparams
some minor changes
2007-07-22
refactor, removed all fall back behaviour in IO.
floatio
- refactor: let the caller of str2int set the prefix
floatconvert
- refactor: simplify _outeng
no _out* has fall back behaviour any more, it returns an error
instead
testfloatnum
- removed a few test cases, that tested fall back behavior
2007-07-21
refactor
buffer overrun checking completed
floatio:
- added comments
- refactor: new type t_buffer, to allow overflow checking
- refactor: make all digit sequence handlers use t_buffer
- refactor: encode* was called only once, include the code in caller
- refactor: all *2str functions report error in case of buffer overflow
- refactor: get all the t_ioparams stuff out of the core of floatnum
floatconvert:
- refactor: adjust _emptytokens to handle t_buffer
- refactor: all output handlers report buffer overflow
- refactor: get all the t_ioparams stuff out of the core of floatnum
- refactor: get all fallback logic out of the core of floatnum
- bug fix: fixed memory leak in error path
hmath:
- refactor: doFormat uses t_buffer now
testfloatnum:
- adapt test cases to new interface
2007-07-20
refactor, comment
hmath
- use IO_FLAG_SHOW_BASE and IO_FLAG_SHOW_EXPBASE
- cattokens needs no buffer for sign encoding
floatio
- started commenting
- refactor: rename leading0 to leadingSignDigits in t_seq_desc
- refactor: Flags IO_FLAG_SHOW_NONDEC_BASE and IO_FLAG_SHOW_NONDEC_EXPBASE
replaced by saner IO_FLAG_SHOW_BASE and IO_FLAG_SHOW_EXPBASE
- refactor: make _significantdigits public, as it can be used in floatconvert
- refactor: input: restrict knowledge about sign characters to the parser only
- refactor: input: do not separate complement information from sign information
- refactor: input: restrict knowledge about basetags to the parser only
- refactor: input: sanitize maxdigits handling
- refactor: input: reduce dependancies, only the parser knows about t_ioparams
- refactor: output: restrict knowledge about sign characters to outermost cattokens only
- refactor: output: do not separate complement information from sign information
- refactor: output: restrict knowledge about basetags to the cattokens only
- refactor: output: pass base of exponent directly to float_out
- refactor: started implementing buffer overrun checking
floatconvert
- refactor: rename leading0 to leadingSignDigits in t_seq_desc
- refactor: use _significantdigits
2007-07-18
bug fix
provide an INSTALL file for those curious enough to try floatnum
floatconvert
- to convert really huge numbers, a constant in _lg2 needs to be
specified more precisely.
2007-07-17
got floatnum running under SpeedCrunch!
first bug fixes
hmath
- bug fix: operator>> should use float_shr, not float_shl
- get rid of HMATH_COMPARE_PREC
- for 80 digits precision, a working precision of 83 is sufficient
- general mode for non-decimal bases has now fractions equivalent
to HMATH_MAX_SHOWN
floatgamma
- _lngamma, _gamma return NaN in all error cases
- refactor: rename _lngamma_ext to _ln_gamma_prim
- refactor: lngamma_prim: separate pos x handling from neg x
- refactor: remove special cases x==1, x==2 from _lngamma_prim
- refactor: _pochhammer: leave the pole detection to the subroutines.
floattrig
- bug fix: _trigreduce returned a non-zero value in case of error
floatio
- new flags IO_FLAG_SHOW_NONDEC_BASE and IO_FLAG_SHOW_NONDEC_EXPBASE
2007-07-16
finally the two-valued factorial of SpeedCrunch, which is
a rephrasing of the Pochhammer Symbol, which in turn can be
written as a quotient of two Gamma function values, is fully
supported.
That means, that the quotient does not necessarily overflow,
even if one of the Gamma values does, that cancellation
of poles is handled properly, and that neither the difference
of the arguments to the factorial nor any argument itself
need to be an integer.
Started merging the latest changes from 0.8
hmath
- added extra tests to nCr and nPr, because the redefined
factorial would otherwise interpolate and extend both functions
(e.g. nPr(-21, 2) == residue(Gamma(-22))/residue(Gamma(-20)) == 462
and not NaN)
- added Gamma and lnGamma
- added logic operators
- added formatEngineering
- added constant phi and limited it to 130 digits
- added gcd
- added cot and friends
- added min/max
- added statistic functions
floatgamma
- evaluation of rising Pochhammer symbol, Gamma and ln Gamma are
based on a primitive, that allows handling of poles, and
negative values.
- ln Gamma is extended to negative arguments by redefining it
as ln abs Gamma x
2007-07-15
CMakeList.txt
- added all floatnum files
- added compiler switch -g to allow easy debugging
testfloatnum
- typo: missing _ in _FLOATNUMTEST
- rewrote the test suite for the Pochhammer symbol
floatgamma
- refactoring: use _sinpix in several places
- in _pochhammer: corrected handling of poles in the denominator
- some bug fixes in _pochhammer
- replaced a few EXACT parameters, that lead to trouble with
huge numbers
- in _ofs: no need to do evaluations using floatnum
floattrig
- moved evaluation of sin(pi*x) here
floatcommon
- this was needed in several places: float_isodd
2007-07-14
Added all higher functions to hmath
The Pochhammer symbol needs some testing
hmath
- refactoring: removed duplicated code
- pi value used from floatconst
- replaced higher mathematical functions by floatnum's functions
- introduced inverse functions of the hyperbolic functions
floathmath
- added float_raisei
- added float_lg
- added float_pochhammer
- bug fix in float_tan
floattrig
- bug fix in trigreduce
floatgamma
- removed a memory leak
- generalized the pochhammer symbol
2007-07-13
started reworking hmath, got a first version running.
The testhmath suite is almost completely passed. The few failures are
all explicable and do not really indicate a problem.
testfloatnum
- compile an empty program if _FLOATNUMTEST is not set
CMakeList.txt
- added some floatnum sources to testmath
- added floatipower to the sources list of testfloatnum
hmath
- rewritten to use floatnum
floatnum
- float_setsignificand, float_setscientific use const char* now
floatconvert
- some const modifiers added
- other minor changes
floatio.h
- added missing extern "C" tag
- removed include of floatnum.h, floatio is based on text only
- forked t_tokens into two almost identical structs, one for
input, using const char*, the other for output, using char*
many const modifiers needed for this
floatipower.c floatipower.h
- new files that contain all the power functions not using logarithms
This way, conversion does not include all the high math stuff
floatpower.c floatpower.h
- moved all integer based power functions to floatipower
2007-07-12
first v0.01 release of floatnum
two io routines linking floatnums io to external parsers added
last issues connected to two's complement io are resolved.
only error and corner cases are not tested yet.
2007-07-10
again a rework of the io
floatlong.h floatlong.c
- removed unused _longintneg function
2007-07-05
major rework of io routines. Mostly floatio and floatconvert are effected,
but there are smaller improvements on other files as well. Base 10 output
seems to run, base 16 output has still to be tested thoroughly.
floatconfig.h
- fixed a bug due to typo
---
testfloatnum.c
- added new testcases
---
floatlong
- refactor: move t_longint and basic procedures here
---
2007-06-21
complete new input parser capable of parsing inputs of base 2, 8, 10 and 16
floatio.c floatio.h
- added new input parser
---
floatconvert.c floatconvert.h
- new conversion routines float_parse and float_str2num converting input
strings to floatnum, bases 2, 8, 10 and 16 supported
---
testfloatnum.c
- added test cases of float_str2num
- added convergence tests of gamma series
2007-06-17
started input parsing
floatio.c floatio.h
- added several input parsing routines
2007-06-16
floatlogic.c
- removed unused _canconvert
- bug fix in floatnum2logic: removed memory leak
- bug fix in floatnum2logic: crashed when 0 < |x| < 1
---
floatio.c floatio.h
- new files containing all IO related stuff like character
handling
- refactor: output routine significand2ascii
---
floatnum.c
- refactor: moving most of float_getsignificand to floatio
- deleted some unnecessary stuff in float_setsignificand
2007-06-15
Logic functions seem to be ok. I'm sure, some corner cases
are still lurking, that will make an operation fail, but on the whole...
Gamma function and family seem to be ok as well
Next, special output routines will be addressed. A standard
output routine is already available (for bases 10, 16, 8 and 2)
---
CMakeList.txt
- updated dependencies
---
number.c
- fixed a bug, bc_int2num could not convert the most negative integer
---
testfloatnum.c
- added test_pochhammer
- added test_lngamma
- added test_gamma
- added test_gammaint
- added test_gamma0_5
- added test_longint2floatnum;
- added test_floatnum2logic
- added test_logic2floatnum
- added test_not
- added test_and
- added test_or
- added test_xor
- added test_shl
- added test_shr
---
floatgamma.c
- bug fixed in _pochhammer
- added special case treatment in _lngamma for x == 1, x == 2
- adjusted working precision in _lngamma, _gamma
---
floatconst.c
- fixed a bug
- added constant cUnsignedBound
---
floatconvert.c floatconvert.h
- added reverse conversion _longint2floatnum
- added support for zero in floatnum2longint
---
floatlong.h floatlong.c
- added _longshl
---
floatcommon.h floatcommon.c
- boolean result type is char
---
floatlogic.c floatlogic.h
- new files containing a draft of logical operations
---
floathmath.c floathmath.h
- added float_not, float_and, float_or, float_xor, float_shr, float_shl
2007-06-14
conversion seems to run, at least in ordinary cases.
added a draft version of factorial computation.
testfloatnum.c
- added test cases to test_stdconvert
- added test_lngammaseries
---
floatconst.c
- added Bernoulli numbers to constant list
- added constant 12 and -20 to constant list
- extended precision of constants to 120
- added constant ln(sqrt(2*pi))
- added constant sqrt(pi)
---
floatseries.h floatseries.c
- added lngammaseries
---
floatgamma.c floatgamma.h
- draft version of _gamma, _lngamma, _pochhammer, _gammaint, _gamma0_5
---
floattrig.c floattrig.h
- boolean result type char
---
floatexp.h floatexp.c
- boolean result type char
- added new function _power10
---
floatnum.h floatnum.c
- boolean result type char
- removed test_floatnum declaration
---
floatpower.h floatpower.c
- boolean result type char
---
floathmath.h floathmath.c
- boolean result type char
- added functions float_power10, float_gamma
---
floatcommon.c floatcommon.h
- added float_asinteger
2007-06-13
the conversion algorithm is in many parts ok, a lot of test cases
are handled correctly. Some tests are still outstanding
ChangeLog (this file)
- reverted the order of entries. It is better to have the latest
at the beginning
---
floatconvert.c floatconvert.h
- renamed u_array to t_uarray
- renamed t_bigint to t_longint
- reworked interfaces to use t_longint
- introduced a new struct t_convert to hold an intermediate conversion state
- reworked interfaces to use t_convert
- rewrote many parts
- bug fix: half size of UARRAYLG is big enough, because a char contains 2
hexdigits
- fixed several bugs in the conversion algotithm
---
floatconfig.h
- adding a comment explaining the limit on BITS_IN_EXP
---
testfloatnum
- adapted test cases to interface changes
- added test cases for _stdconvert
2007-06-12
tying together the parts to implement a draft conversion algorithm
For the moment, to hex only, and to intermediate data structures
CMakeList.txt
- updated dependencies
---
floatconst.h floatconst.c
- added constant 16 for conversion
---
floatconvert.h floatconvert.c
- declared t_bigint, to be used later
- implemented a draft version of a hex conversion algorithm, has to be tested
---
floatgamma.c floatgamma.h
- introduced empty templates to the code base.
After all, that's where all started from.
2007-06-11
testfloatnum.c
- added test cases to _raiseposi
- added test_raisei;
---
floatpower.c
- fixed a bug in _raiseposi
- adapting scale in _raiseposi, so that the error
is always < 1e-99 for 100-digit computations
---
all
- renamed arcsinh to arsinh, arccosh to arcosh and arctanh to artanh
2007-06-10
radix conversion is on its way.
floatconfig.h
- introduced a define MATHPRECISION, that controls to
what maximum precision a higher mathematical function
can be called
---
floatconvert.c
- changed return type of _bitsubstr to unsigned, so we
can deal with more than one digit at a time.
- introduced type u_array
- fixed typo in name '_floatnum2uarray'
fixed bug in _floatnum2array
- replaced all uarray2* by a generic uarray2str
---
floattrig.c
- using MATHPRECISION
---
floatcommon.c
- using MATHPRECISION
---
floatconst.h
- removed FLOATMAXPRECISION. MATHPRECISION has taken over
its role
---
floatpower.h floatpower.c
- modified the interface of _raiseposi slightly
---
testfloatnum
- added test_floatnum2uarray
- added test_uarray2str
- added first testcases for _raiseposi, more pending
2007-06-09
floatlong.c floatlong.h
- added an overflow-checking addition routine _checkadd
- added an overflow-checking multiplication routine _checkmul
- moving _findfirstbit here and make it public
---
floatconvert.c
- using floatlong's version of _findfirstbit
---
floatpower.c
- removed _lgi
- use floatlong for exponent calculations
---
floatnum.c
- clean up: using _checkadd in float_setscientific
---
testfloatnum.c
- activated somehow dropped test_setzero, test_getsign, test_getlength,
test_setscientific tests
- added test_checkadd
- added test_checkmul
- added test_longarrayadd
- added test_longarraymul
---
floatconfig.h
- removed define FLOATMATHTEST. FLOATNUMTEST is enough
---
all
- spelling: mathma.. replaced by mathema..
2007-06-08
heading towards the missing radix conversion routines. Necessary, before
a hmath version can be installed. As a side effect, the power operation
is developed
CMakeList.txt
- updated dependencies
---
floatlong.h floatlong.c
- introduced these two files to the code base. They contain a portable
long integer arithmetic
---
testfloatnum.c
- added a final message
- added test cases for _longadd, _longmul, _longshr
---
floatnum.c
- using floatlong's _long* functions in float_setscientific instead of
special bit pattern based integer overflow detection
---
floatexp.c
- in _exp: cleaned up the setting of the exponent and added a safety check
---
floatcommon.c floatcommon.h
- introduced a new function float_isinteger
- using floatlong's _long* functions in leadingdigits. Should now
return correct results without exception
---
floatpower.h floatpower.c
- introduced these two files to the code base. They contain the basic
routines for the power operation
---
floatconvert.h floatconvert.c
- introduced these two files to the code base. They contain the basic
procedures for converting a number to a char array based on radix 16, 8, 2
---
floathmath.h
- added comments
---
all
- whitespace removing
- replaced 'unsigned int' by 'unsigned'
2007-06-07
The first target testfloatnum is compiling and running, and all tests are
passed.
CMakeList.txt
- updated dependencies
---
floatconfig.h:
- moved configuration settings into a separate file, floatconfig.h.
- introduced a define BITS_IN_EXP, the number of bits available for
the exponent (including the sign of the exponent) in a floatnum. This
definition is more basic than EXPMAX, and useful when implementing exponentiation.
---
floatnum.h
- introduced a new error code, FLOAT_UNSTABLE. It indicates a situation where
the mathematically correct result cannot be evaluated due to limitations
of the algorithm. Used with higher mathematical functions only, not with the
basic operations in floatnum.
- updated many comments to reflect the latest state and to improve the style.
---
floatnum.c
- removed defines that are handled in floatconfig.h now
- fixed float_sqrt and float_round, where a spurious result value could be returned.
- declared _movepoint explicitly static void
- introduced a new function float_getdigit to access a single digit in
a significand.
- removed superfluous variables
- updated and introduced new comments
---
testfloatnum.c
- removed defines that are handled in floatconfig.h now
- removed function declarations that were leftovers from changes
- removed conflicting definition of float_error
- correct initialization of floatnum module
- made tests verbous
- extended tests to higher mathematical functions
---
floatcommon.h, floatcommon.c
- introduced these two files to the code base. They contain declarations
and definitions for helpers and convenience functions, used throughout
the evaluation of higher mathematical functions
---
floatconst.h, floatconst.c
- introduced these two files to the code base. They contain declarations
and definitions of constants used throughout the evaluation of higher
mathematical functions
---
floatseries.h, floatseries.c
- introduced these two files to the code base. They contain declarations
and definitions of routines for evaluating Taylor series of cos x - 1,
cosh x - 1, arctan x, arctanh x, used throughout the evaluation of higher
mathematical functions
---
floathmath.h floathmath.c
- introduced these two files to the code base. They contain declarations
and definitions of the top level calls to higher mathematical functions
---
floatlog.h floatlog.c
- introduced these two files to the code base. They contain the basic
routines for all functions related to the logarithm
---
floatexp.h floatexp.c
- introduced these two files to the code base. They contain the basic
routines for all functions related to the exponential function
---
floattrig.h floattrig.c
- introduced these two files to the code base. They contain the basic
routines for all functions related to the trigonometric functions and
their inverses
---
hmath.h
- corrected comments related to hyperbolic functions. Hyperbolic functions
are not based on angles, so they do not depend on any degree setting
2007-06-03
number.c:
- fixed a bug in bc's number.c, function bc-divide, that causes trouble when
overwriting the dividend while dividing by 1. floatnum must have this fixed.
|