File CHANGES.TXT

ARIBAS interpreter for Arithmetic, V1.64, January 2010
Copyright (C) O.Forster 1996-2010 <forster@mathematik.uni-muenchen.de>
WWW:	http://www.mathematik.uni-muenchen.de/~forster

Changes from former versions of ARIBAS to the present version:

V 1.64 (Jan. 2010)
- bug fix in trigonometric functions with high floating point precision
- raised maximal floating point precision to 5120 bits

V 1.60 to 1.63 (Feb. 2008)
-  bug fix in open_write
-  corrected behavior of function round
-  fixed bug occuring in some gf2n_ and gf2X_ function on bigendian architectures
-  bug fix in function gf2X_mod

V 1.50 to 1.60 (Aug. 2007)
-   direct support for polynomial arithmetic over GF(2),
      functions gf2X_mult, gf2X_square, gf2X_divide, gf2X_div,
      gf2X_mod, gf2X_gcd, gf2X_modpower, gf2X_primetest
-   bug fix in handling of -pi (previous version changed constant pi)
-   bug fix in division of integer vectors with negative coefficients
-   for loop can now handle >= 2**32 iterations
-   fixed bug which occurred in V 1.50 while parsing certain
    parenthized expressions

V 1.40 to V1.50 (Aug. 2004)
-   direct support for GF(2**n); data type gf2nint,
      functions gf2n_init, gf2n_fieldpol, gf2n_degree,
      gf2n_trace, max_gf2nsize
-   new function gfp_sqrt
-   removed a bug in integer division
-   new function ec_factorize
-   improved performance of qs_factorize (less unsuccessfull
        factorizations, thanks to Lew Baxter for testing)

V1.30 (March 2002) to V1.40 (June 2003)
-   simultaneous assignments
        (x1,x2,...,xn) := (a1,a2,...,an)
-   new functions 
        divide, bit_count,
        set_workdir, get_workdir, gmtime,
        stack2string, stack_arraypush, string_scan,
        realloc, binsearch
-   some bug fixes (regarding qs_factorize and
	    handling of special variable and value arguments
        in user defined functions)

V1.20b (June 2001) to V1.30 (March 2002)
-   vector operations:
        vec + vec, lambda * vec, vec * lambda,
        intvec mod N, intvec div N, gcd(intvec)
-   some small bugfixes
-   fixed bug occuring in special situations when reading comments

V1.14 (April 1999) to V1.20b (June 2001)
-   flush(transcript)
-   floating point precision can be set up to 4096 bits (over
    1200 decimal places); former limit was 192 bits
-   bugfix in function mem_byteswap()
-   make_unbound(user) unbinds all user defined functions
-   new function max_intsize(); returns maximal number 
    of decimal places of an integer

V1.08b (Nov. 1997) to V1.14 (April 1999)
-   since V1.13 there is now a (preliminary) version of ARIBAS
    for MS-Windows 95/98/NT
-   function load in UNIX versions of ARIBAS now accepts also 
    .ari files with DOS line endings
-   silent version of functions rho_factorize, cf_factorize,
    qs_factorize, next_prime with last argument 0.
-   new builtin function next_prime 
    next_prime(x) calculates the smallest prime p >= x
-   new builtin function qs_factorize (factorization with
    quadratic sieve method, faster than cf_factorize; on a 
    Pentium PC 233MHz, the seventh Fermat numbers is factorized
    in a few seconds)
-   continue statement in for and while loops (works as in C)

V1.07 (April 1997) to V1.08b (Nov. 1997)

-   fixed bug in factorial(n) for n=0,1
-   readln() can now read big integers which extend over several lines
-   new command line option -b (batch mode):
        aribas -b tt.scr
    starts ARIBAS, loads file "tt.scr" which is supposed to contain
    ARIBAS code, executes it, and then exits.

V1.00 (Sep. 1996) to V1.07 (April 1997)

-   new functions version() and max_arraysize()
-   function protocol() renamed to transcript()
-   new command line options -h (help path) and -p (ari path)
-   (versions for 80[3-5]86 processor) small speedup in big integer arithmetic
-   (MSDOS version) changed memory management for function system()
-   bugfix in function random()

(************************** EOF ******************************)