# $Id$
Bug numbers refer to the BTS at http://pari.math.u-bordeaux.fr/Bugs/

Done for version 2.5.0 (released 31/05/2011):

  Fixed
BA  1- DESTDIR did not work. [#1194]
BA  2- Improve Darwin shared library support.
LGr 3- bnfinit -> oo-loop if class number is divisible by 27449 [#1197]
BA  4- [gp2c] error("foo") did not work.
    5- wrong generators in quadclassunit(D > 0) [ fix qfr3_pow / qfr5_pow ] [#1195]
    6- setrand(45);quadclassunit(185477) -> SEGV [#1200]
    7- fix derivnum for large arguments [#1201]
    8- "#if READL" always evaluated to "FALSE" in gprc [#1202]
    9- matsnf(non-square matrix, 4) --> SEGV [#1208]
HC 10- fix p-adic Gamma inefficiency for small p: gamma(O(2^20))->no answer

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Done for version 2.4.4 (released 29/04/2011):

  Fixed
BA  1- FpX_gcd did not garbage-collect for large input.
BA  2- 'make all' failed to build the documentation.
    3- typo in polgalois() when new_galois_format is 0 : returned [3,1,2]
       instead of [3,1,1] for C_3
    4- rnfkummer was using too much stack.
    5- nfgaloisapply did not work on extended ideals.
    6- Strprintf and printf0 prototypes were wrong.
    7- Mod(4*x+1,Mod(1,2)*x) -> SEGV [#1116]
    8- gcd(t_INTMOD, t_FRAC) -> SEGV
    9- qfjacobi failed on a matrix of zeros (or in presence of 0s of large
       exponent) [#990]
   10- t_SER / t_SER where denominator has 0 leading term -> SEGV [#1120]
   11- (t_COMPLEX of t_INTMOD) + t_PADIC -> SEGV [#1121]
   12- (t_QUAD of t_INTMOD) + t_PADIC -> SEGV [#1122]
LGr13- bnfinit: various performance improvement.
BA 14- several instance of random garbage collection were broken.
BA 15- [install]: support for D0,G, and D"str",s, in prototype code.
BA 16- [gp2c]: foo(x:mp)=... did not work.
   17- when the divisor's leading term was 0, RgX_divrem(,,&rem) created rem
       before the quotient (invalidating the use of cgiv() suggested in the
       documentation) [#1129]
   18- gadd(t_COMPLEX of t_REAL, t_QUAD = 0) -> SEGV [#1131]
   19- log(11+11^2+O(11^3)) -> bug in log_p [#1136]
BA 20- (ffgen(ffinit(2,10))*0)^-1 -> oo-loop
BA 21- ispseudoprime(-3,0) != ispseudoprime(-3,1)
BA 22- text form of a closure now include the closure context.
BA 23- reading compressed files could corrupt the stack.
BA 24- t_CLOSURE==t_CLOSURE and t_CLOSURE===t_CLOSURE did not work.
   25- gequal1(0.*x + 1) returned 0
   26- ispower((100!)^2) -> 0 [#1155] Affected pure powers divisible by more
       than 22 of the 26 primes <= 101. (introduced in 2.4.3)
BA 27- LLL could fail (rarely) due to insufficient precision [#1154]
   28- rnfisnorminit(y,x^2-Mod(2+y,y)): polynomial not in Z[X] in nfinit [#1157]
   29- factorff failed to merge identical irreducible factors [#1159]
BA 30- ploth(x=0,1,0,"Complex|Recursive") -> SEGV [#1163]
   31- mathnf(dim > 8) could return a non-HNF upper triangular matrix [#1153]
BA 32- forsubgroup() did not handle break/next/return.
   33- modular resultant, e.g. rnfequation(polcyclo(17,y), x+y^6,1) [#1151]
   34- gequal1(x) / gequalm1(x) did not implement gequal(x, 1) for t_COL/t_MAT
   35- matsize(vecextract(M,0,y)) -> [;] instead of matrix with 0 and the
       requested number of columns [#1170]
   36- readline completion : allow commas ',' in directory names [#105]
   37- List() == List() returned 0.
   38- [Configure] C_INCLUDE_PATH was not used when looking for libgmp
BA 39- polroots(1) returned [] instead of []~.
   40- a succession of <C-C> in breakloop context would reinitialize timer
   41- matsnf(matdiagonal([x,0])) --> [x,0] instead of [0,x]
BA 42- qfbred(large discriminant, small prec): precision too low in truncr [#1173]
PMo43- erfc(large x) was slow and inaccurate [#364]
   44- erfc() : allow t_COMPLEX inputs
   45- stack corruption in ifac_decomp_break() [#1115]
   46- factor(0) returned 0^1, but factorint(0) raised an error. Allow
       factorint(0), core(0), coredisc(0), etc.
   47- ellwp, ellzeta, ellsigma only used realprecision, instead of the input
       precision [#1184]
BA 48- tuning for 32bit and 64bit, with or without GMP5.
   49- idealintersect(K, 1/2,1/2) -> 1/4 instead of 1/2 [#1192]

  Changed
LGr1- universal constants (gen_0,gen_1, etc.) are now read-only
   2- qfjacobi : sort eigenvalues by increasing order
   3- [libpari] gtoser : add a third precdl argument.
   4- [libpari] swap the last 2 arguments of ellwp0 (prec and precdl)
   5- ellwp now outputs a number of terms equal to the default seriesprecision,
      instead of *twice* that number.
   6- Ser, ellwp, taylor, elltaniyama: add an optional 'precdl' argument (by
      default equal to 'seriesprecision'). One no longer has to rely on (and
      change locally) a global variable to handle conversion to power series.
LGr7- add information to the version() output [backward compatible if version()
      used as recommended in the documentation] [#1130]
   8- O(x^0) is now printed as is, not as O(1). t_SER and t_PADIC are now
      treated in the same way.
   9- inverse trigonometric functions (acos,asin,atan,acosh,asinh,atanh):
      values on the branch cuts changed to conform to standards, e.g.
      "implementations shall map a cut so the function is continuous as the cut
      is approached coming around the finite endpoint of the cut in a counter
      clockwise direction." (ISO C99) [initial patch Richard Kreckel, #1084]
BA 10-[libpari] nf_get_TrInv renamed to nf_get_diff
   11- Qfb() : forbid square discriminants (-> corrupted objects [#1145])
   12- allow gequal1 / gequalm1 to return 'true' for t_SER
   13- follow-up to (2.4.3, C104) [#1156]:
       - stop using the undocumented feature Z_factor_limit(x,1) = Z_factor(x).
       - abide by the "lim" parameter in all cases :
          factor(100,1 or 2) -> [10; 2] (was [2,2; 5,2])
   14- include GP defaults in the function description system + use standard
       hashtables instead of ad hoc types
   15- cleanup and document term_color() and term_get_color()
BA 16- prototype of summations functions now is fun(void *E, GEN call(void*, GEN),...)
BA 17- traversesubgroups renamed to forsubgroup.
   18- t_QFR with the same coefficients but different distance component are
       now tested equal by == (they are still different according to ===).
   19- bnfinit no longer outputs a warning when fundamental units couldn't be
       computed (annoying and rather useless) [#1166]
   20-[Configure] no longer support a link to an uninstalled readline library
      in PARI toplevel
   21- thue() no longer outputs a Warning when the result is conditional on
       the GRH.
   22- [libpari] rename TIMER -> timer_delay, TIMERread -> timer_get,
       TIMERstart -> timer_start, msgTIMER -> timer_printf
BA 23- polrootsff now returns a t_COL
   24- default(a, b) used to return "b". Now returns gnil
   25- [libpari] remove allocatemem0 [make it private to GP], write allocatemem
   26- matdet() : allow Guass-Bareiss to develop somewhat wrt. rows/columns
       [ cf test O1 in Lewis-Wester's bench ]
   27- addprimes() now includes its argument as-is (used to take gcds,
       making an insertion linear in the table size instead of O(1); filling
       an inially empty table was quadratic in the final table size). They
       must be true primes, otherwise number theoretic routines may return
       wrong values. [#322]
   28- rename fprintferr -> err_printf, flusherr -> err_flush
   29- ellpow (CM case). Try to determine the discriminant of the
       endomorphism ring first [#1186]

  Added
BA  1- trap keyword "syntaxer" to trap syntax error from eval.
BA  2- Flx_shift now support negative shift
BA  3- [gp2c] add description _(_) for closure evaluation
BA  4- PARI function FpX_halfgcd() and subquadratic FpX_gcd(), FpX_extgcd()
BA  5- PARI function Flx_halfgcd() and subquadratic Flx_gcd(), Flx_extgcd()
BA  6- random() now allow to draw random polynomials
BA  7- PARI functions pol_0, pol0_Flx, pol0_F2x
VB  8- [mingw] --datadir=@ option to Configure
BA  9- PARI functions FlxqX_div
   10- PARI function ismpzero
BA 11- PARI function bnr_get_clgp
   12- new default 'histfile' (to save readline history in between sessions)
   13- PARI function cmp_universal
BA 14- [gp2c] add descriptions to support intnum/suminf type functions.
BA 15- function diffop
   16- PARI function zero_Flm_copy
   17- PARI function mantissa_real
BA 18- PARI functions F2v_clear, F2m_clear, Z_to_F2x, F2x_set, F2x_clear, F2x_flip, F2x_coeff
BA 19- PARI functions F2xq_sqrt, F2x_deriv
BA 20- PARI functions FpXQ_trace, Flxq_trace, F2xq_trace
BA 21- PARI functions FpM_det, Flm_det, F2m_det

  Removed
BA  1- PARI function CM_CardEFp (ellsea script deprecated).

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Done for version 2.4.3 (released 08/10/2010):

  Fixed
    1- \h m-n no longer worked
BA  2- some error messages displayed internal token names (e.g for '1)
    3- named colors in 'graphcolormap' were not portable across graphic drivers
    4- setunion(Set(),Set()) -> SEGV [#714]
DE  5- [Solaris] pari.desc won't compile [#715]
    6- 822.pm broke the formatting of inline program examples
    7- memory leaks in pari_close_opts [ freeep(), parser stacks, some
       defaults... ]. FIXME: calls to getpwuid introduce (small) leaks
    8- ZV_Z_mul might not reset clonebit.
    9- nfeltreducemodpr --> gerepile errors [#716]. Also SEGV for residue
       degre > 1 (when p does not divide the index)
JD 10- write1("/dev/null", foo);1 --> extra newline
JD 11- is_dir_stat() gave incorrect results on special files (pipes,...)
JD 12- read() would block on pipes
   13- nfeltreducemodpr used different canonical representatives than all other
       modpr routines [#719]
   14- rnfconductor(,,1) did not recognize some abelian extensions [#718]
       (when the discriminant of a defining polynomial for the absolute number
       field could not be factored). Related to Changed-4.
   15- polgalois could confuse S_11 and F_110(11) if a certain polynomial
       discriminant could not be fully factored. Related to Changed-4.
   16- a = 0./x; a+a --> 0.
BA 17- [from 2.4.2] use of pointers in recursive expressions [#717]
BA 18- [from 2.4.2] trap() did not restore evaluator states [#722]
   19- simplifications missed when adding t_RFRAC [#721]
   20- gtofp(1 + 0*I, prec) should not return 1. + 0.*I but 1.
       gtofp(0 + (tiny real) * I, prec) should make sure the real part is
       converted to a zero of sufficiently small exponent (otherwise the
       modulus of the result is zero, because 1E-100 + 0E-28 = 0E-28)
   21- factor(500501^36): impossible assignment I-->S
   22- quadray(-7,6) [or other "impossible conductors"] --> oo loop [#725]
   23- acosh([Pol(1)]) -> gerepile error [#728] (same asin, asinh)
   24- rnf functions sometimes do not reduce their output when relative
       extension has degree 1 [#727]
   25- zetakinit(bnfinit(...)) -> gerepile error
   26- zeta(1-2^31 or 1-2^63) -> SEGV [#730]
BA 27- priority of << had become lower than priority of * (should be =)
   28- last digit of polylogs was wrong
BA 29- missing GC in the leftrightpow functions
   30- bnr=bnrinit(bnfinit(y^2-229,1),5,1); rnfkummer(bnr,,3) --> oo loop
   31- mathnf(a,1) could have negative entries if non-trivial kernel [#741]
   32- [Configure -pg] some variables not written to pari.cfg
   33- missing subfields in nfsubfields [#744]
   34- [readline] <F1> or <M-h> would fail on 'fun()'. Delete trailing
       garbage in help routine.
   35- modular matrix inversion routines used O(n^2) inversions instead of O(n)
   36- print1() + read() or extern() + error/warning ==> missing \n
   37- infinite loop in nffactor [#751]
   38- ellordinate couldn't handle t_FFELTs
   39- eint1(10,10) completely wrong
   40- ispower broken due to 1) uninitialized variable, 2) insufficient
       precision [#916]
   41- compilation fails with yacc unable to handle parse.y [ add explicit
       rule in Makefile to prevent from generating a wrong one ]
   42- duplicate factors in nffactor (easy cases) [#761]
   43- ellisoncurve(e, P) didn't check that "point" P had the correct format
   44- znlog(...,znprimroot(p)) --> SEGV [#765]
   45- wrong signs in bnfisintnorm e.g. [bnfinit(x^3+5), 5] -> x [#767]
   46- x/(x+2) * Mod(1,2) -> Mod(1, 2)*x/(x + 2)  [ should be Mod(1,2) ]
   47- substpol(x/(y^3*x+1),y^2,1) -> x/(y^3*x + 1) [ should be x/(y*x+1) ]
   48- missing sanity checks in rnfbasistoalg()
   49- setrand(1); E=ellinit([0,0,0,1,T^4]*Mod(1,ffinit(2,7,T))) --> "weird
       base ring; can't divide"
   50- nffactor(nfinit(y),x*y)
       nffactor(nfinit(y),x^2*y+1)
       nffactor(nfinit(y),x^2*y)  --> errors or SEGV. [#769]
   51- hilbert(1,1,1) --> oo loop [#770]
   52- nffactor(y^2+1, x+1/2) --> x+1/2 [ not integral ]
   53- nffactor(y^2+1,x*y+1) --> x*y + 1 [ not normalized, missing POLMOD ]
   54- nffactor(y^2+1,x^2+2*y+1) --> x^2+2*y+1 [ missing POLMOD ]
   55- ellpow() with CM raised a "division by 0" instead of returning [0] [#777]
   56- ellpow(e,z,2^20*I) --> norm too large in CM.
   57- p=x^4-x^3+x^2-x+1; tnf = thueinit(p); thue(tnf,1) -> missing solutions
       (could occur whenever p has no real root) [#764]
   58- missing GC in numtoperm()
   59- harmless uninitialized memory read in mulrrz_i()
   60- sigma(..., 2) --> SEGV [ introduced in 2.4.1 ] [#783]
   61- tuning problems in bnfinit
       bnfinit(polredabs(quadhilbert(-2939))) --> oo loop
   62- minor inaccuracy in GMP kernel (affir)
   63- missing GC in gamma / lngamma
   64- obscure SEGV when setting 'log' in a file read from command-line
       as gp foo.gp [#789]
   65- check that moduli are compatible in Rg_to_Fp(t_INTMOD, t_INT)
   66- matsolve([1;2;3],[1,2,4]~) --> [1]~  (should raise an error)
   67- allow polredabs(x in Q[X])
BA 68- [galoisinit] Documentation of "weakly" supersolvable was incorrect.
   69- binomial(t_REAL or t_COMPLEX, k) was very slow for large k
   70- quadclassunit(-3 or -4) --> one extra (obsolete) component.
BA 71- the evaluator did not garbage collect huge expressions [#713]
BA 72- changing a function in use caused a crash [#784]
   73- vecsort() would not check the type of element in 1-dimensional vector
       [ e.g. vecsort([I] now raises an error ]
   74- RgX_val would raise a SEGV on t_POLs with exact *non-rational* 0
       leading coeff, e.g. Mod(0,2)*x
   75- SEGV (stack corruption) in idealval(K, rational number, pr) [#808]
   76- All the charpoly(x,v) variants failed when v > variable(x)
BA 77- conjvec did not support FFELTs.
   78- memory leak in t_LIST handling
   79- polrootsmod(,1) could return an incomplete result [when 0 was a root
       or there was a double root], e.g. polrootsmod(x^3 + x^2 + 21*x,23,1)
IM 80- lngamma(x) gave incorrect results (off by 2*Pi*I) when real(x) was a
       negative half-integer.
   81- trying to write to a binary file and failing leaked a file descriptor
   82- long-standing stack corruption when <C-C> pressed during gclone [#458]
   83- idealaddtoone had problems with 0 ideals [#829]
   84- wrong results in qfgaussred() [only totally real case was correct]
   85- uninitialized memory read in divrr (both kernels), would only affect
       the last bit of the result, but could cause SEGV on empty stack.
   86- After A=1/(y + 0.); B=1. + O(y); all of A*B, A/B, B/A caused an error
   87- wrong result in ellap(E, 2) if bad reduction
   88- 0.*x+1 == 1 --> 1 but 0.*x^2+x == x  --> 0 [ latter is now 1 ]
   89- ffinit(2,30) --> division by 0 [typo in FlxY_Flx_div], [#859]
   90- when unexpected exception/signal raised (^C, stack overflow, out of
       memory), == would cancel it and silently return wrong value [#329]
   91- reading recently freed memory in popinfile()
   92- contfrac(1,[],-1) ==> SEGV [#865]
   93- missing type checks in bnrconductorofchar [#867], nfhilbert [#868]
   94- unstable evaluation of tan(t_COMPLEX); at \p9, try
       ploth(t=-0.1,0.1,abs(tan(10*I+t)))
BA 95- Vecsmall([1,1])^-1 returned garbage
   96- was not handling nffactor(*huge degree* nf, *non-monic* t_POL) properly
       --> wrong result and possible memory corruption [#870]
   97- polredabs: excessive stack usage [#872]
   98- besselj(0., 0) --> error [#880]
   99- polred didn't accept non-monic inputs
  100- factor(..., 10^100) ==> "overflow in t_INT-->long conversion" [#855]
IZ101- missing `const's in groupid.c cause massive copying to stack.
  102- [native kernel only] int_normalize() didn't treat properly inputs
       of lgefint 2
  103- gerepile errors in quadclassunit(13).gen or .cyc
  104- from_Kronecker & FpXQX_from_Kronecker returned polynomials with invalid
       first codeword [ unused bits not reset to 0 ]
  105- gred_rfrac_simple(n,d) [ basic constructor for t_RFRAC n/d ] did not
       cater for d = constant polynomial
  106- RgX_extgcd created invalid objects in trivial cases [ gen_0 used by
       RgX-only routines ]
BA107- Fix detection for ppc64 running in 32bit mode.
  108- Fp_div used remii instead of modii: Mod(3,5)-1/3 -> Mod(0, 5) [#893]
  109- trying to prevent accuracy problems in thue() caused oo loop
  110- wrong results in matsnf with t_POL entries [#901]
       minpoly(matrix(4,4,i,j,i/j)) -> x^3 - 4*x^2 + x
  111- thue(thueinit(x^3-100,1), 25) -> "Short continued fraction"
  112- thue(thueinit(x^3-48, 320) -> missing solutions (similar problem
       when LHS is a polynomial of degree d and RHS includes a d-th power).
  113- completely wrong results in nfsnf
  114- intformal(A,Y) -> 1/2*A^2
  115- uninitialized memory read in sin / cos / exp (possibly fed random
       final digits to the algorithm, normally truncated away before returning)
  116- nffactor(): SEGV if "relifting" was necessary (very rare)
  117- nffactor(): obscure "no factor" bug [when trial division found and
       extracted *non monic* factors], #930
  118- incorrect output type in FpX_factor [#933]
  119- possible stack corruption in polylog() + missing imaginary part.
  120- incorrect final simplification in idealtwoelt(nf, x, y).
       Eg: idealtwoelt(nfinit(x^2+23), 3, 6) -> [0, 0]~
BA121- listsort() could return a pointer referencing freed memory: make it
       return void
BA122- src/kernel/ix86/asm0.h had an incorrect #line entry.
  123- rnfkummer could miss some fields ( bug in rnfnormgroup() ), e.g.
       bnrM = bnrinit(bnfinit(X^4 - 34*X^2 + 1189,1), 5, 1);
       rnfkummer(bnrM, matdiagonal([3,1,1]))
  124- nfgaloisconj(y,1) -> invalid object
  125- weird error message in ellpointtoz on loss of accuracy [#962]
  126- ^C before inital prompt appears ==> SEGV [#965]
BA127- ellsea(,,1) did not work
BA128- COMPAT file was missing some functions renamed in 2.3:
       flisexpr, flisseq -> gp_read_str; lisGEN  -> gp_read_stream
AF129- fix for documentation of polcoeff, extern and readvec
BA130- 1/matrix(2,2,i,j,Mod(0, 2)) -> SEGV
  131- wrong results for qfminim(A,,,2) if minimal norm < 1. E.g. A = matid(2)/2
BA132- besselk(0.001,1) -> SEGV
BA133- galoisinit(x).gen was not a t_VEC
  134- lngamma(10^1000) -> precision too low in truncr. Same for psi()
  135- nffactor could segfault. [#979]
  136- Mod(10,37)^10000000000 -> SEGV
  137- missing Flx_renormalize at the end of Flx_mulspec_basecase() &
       Flx_sqrspec_basecase() [#984]
  138- after \y0, [1,x]*[0,0]~ returned 0 t_INT, which should be a t_POL [#983]
  139- isprime(x) possibly incorrect on 64 bit machines when 10^15 < x < 2^64
LM140- fix for the documentation of deriv, intformal, poldisc and taylor.
JD141- fix for the documentation of install, matsnf and polsturm.
LM142- general documentation spelling fixes.
BA143- detection of exp2 and log2 was broken with C++
BA144- support for shared library on netbsd.
IZ145- [readline] 'foo)' + M-C-b -> infinite loop
  146- factorff / polrootsff : wrong result if t_FFELT coefficients
       in variable 'x
BA147- nfisideal did not reject invalid hnf ideal matrices [#999]
  148- v=vectorv(1);concat([matrix(0,0),v,v,v,v,v,v,v,v,v,v,v,v,v,v,v,v])
       -> SEGV [#1002]
BA149- minpoly([-5,0,-1,1;0,-5,-1,-1;1,1,-5,0;-1,1,0,-5]/3)
       -> wrong result[#994]
BA150- 2/O(5) -> O(5^-1) (should raise an error) [#1009]
  151-  factor(13533236897) -> I/O Warning [#1012]
BA152- usqrtsafe result could be off by one.
BA153- faster algorithm for nfgaloisapply.
  154- bnfisnorm: output could have constant t_POL(MAXVARN) components
       instead of t_INT. E.g. bnfisnorm( bnfinit(x^2+10), 1690 )  [#1026]
IZ155- [OS/2] Fix warning about string signedness.
IZ156- [OS/2] Fix symbol checking in Configure.
BA157- upowuu(2,31) returned a wrong result [#1031]
BA158- [from 2.4.2] issquare(x^2+y^2) -> SEGV [#1027]
BA159- [from 2.4.2] pointers to Vecsmall components were broken.
BA160- [from 2.4.2] ellap could return a wrong result. [#1030]
  161- thue() could return wrong results (ZX_Z_normalize() didn't follow its
       documentation) [#1032]
  162- qfminim([1,2;2,1]*1.,,,2) -> SEGV [#1033]
  163- [doc] const_col, const_vec: typo in prototype [#1040]
BA164- precision(0.E100) returned an invalid object.
  165- nbits2nlong, nbits2prec, nbits2nchar : avoid overflow [#1037]
  166- normalization problems in RgX_Rg_add & friends, leading to t_POL
       with sign incorrectly set to 0, e.g. O(2)*x + 1 [#1042]
  167- t_FFELT + t_FRAC -> SEGV (missing case in gadd)
  168- polred(x) -> stack overflow [#1044]
  169- sign of t_POL whose coefficients were all equal to 0 was not always
       set to 0 (RgX_add / RgX_sub) [#1053]
  170- problems with p-adic gamma function :
       - remove extra digit in Morita's algorithm [ e.g. gamma(1/2+O(3^10)) ]
       - better threshold between Morita and Dwork's algorithm
BA171- qfbhclassno(x) was wrong for x<0
  172- subst(t_LIST,x,y) returned a copy without substituting [#969]
       substvec & substpol didn't work on t_LISTs
BA173- RETRY macro was incorrect and could cause crash [#1064]
BA174- factorpadic(,,,1) was actually using ROUND4 [#1064 again]
  175- nfbasis(x^2+1,,factor(-4)) --> oo loop [#1072]
  176- factorpadic: incorrect conversion from ZX to ZpX [#1073]
  177- bnfinit() : possible overflow in double -> long conversion [#1099]
  178- various inconsistencies in polinterpolate() [#1096]
  179- rnfequation() broken when called over the prime field Q [#1079]
BA180- F2x_divrem could call bfffo(0) which is not defined.
  181- remove restriction on forsubgroup / subgrouplist that cyclic factors
       of all p-Sylows should have less than 2^32 / 2^64 elements

  Changed
    1- [libpari] prototype of gpmalloc, gprealloc [ return void * ]
    2- contfrac(x, b): no longer allow (undocumented) type t_MAT for b
BA  3- builtin-in GP functions can now be used as closures.
    4- allow nffactor(t_POL, t_POL), nfroots(t_POL, t_POL) [ necessary when
       nf.disc could not be factored and addprimes() would have to be used ]
    5- removed undocumented interface FpM_mul(x,y,NULL), and analogously for
       FpV_dotproduct, FpV_dotsquare, FpM_FpC_mul
    6- [library] major names cleanup; rename *lots* of functions:
       BSW_* -> BPSW_*
       Z_factor_limit -> Z_factor_until, auxdecomp -> Z_factor_limit
       assmat -> matcompanion
       certifybuchall -> bnfcertify
       derivpol -> RgX_deriv
       discf -> nfdisc, nfdiscf0 -> nfdisc0
       element_div -> nfdiv, element_divmodpr -> nfdivmodpr,
       element_mul -> nfmul, element_mulmodpr -> nfmulmodpr,
       element_pow -> nfpow, element_powmodpr -> nfpowmodpr,
       element_reduce -> nfreduce, element_val -> nfval
       factorback0 -> factorback2
       factorpadic4 -> factorpadic, factorpadic2 now static
       fundunit -> quadunit
       gcmp0 -> gequal0, gcmp1 -> gequal1, gcmp_1 -> gequalm1
       Flx_cmp1->Flx_equal1, F2x_cmp1->F2x_equal1, FF_cmp1->FF_equal1
       FF_cmp0->FF_equal0, FF_cmp_1->FF_equalm1
       ZV_cmp0 -> ZV_equal0
       gener -> znprimroot, ggener -> znprimroot0
       greffe -> RgX_to_ser (+ change prototype)
       hil0 -> hilbert, hilii -> hilbertii [ delete useless hil() ]
       ideal_two_elt* -> idealtwoelt*,
       idealhermite -> idealhnf
       initalg -> nfinit, initalgred -> nfinitred, initalgred2 -> nfinitred2
       initell -> ellinit, smallinitell -> smallellinit
       isfundamental -> Z_isfundamental
       ismonome -> RgX_is_monomial
       isunit -> bnfisunit
       matrixqz -> QM_minors_coprime
       matrixqz2 -> QM_ImZ_hnf
       matrixqz3 -> QM_ImQ_hnf
       minideal -> idealmin and change prototype
       mu -> moebius, gmu -> gmoebius,
       nfhermite -> nfhnf, nfhermitemod -> nfhnfmod, nfsmith -> nfsnf
       phi -> eulerphi, gphi -> geulerphi, phiu -> eulerphiu,
       poldeflate -> RgX_deflate_max
       polymodrecip -> modreverse
       powraw -> qfbpowraw, compraw -> qfbcompraw
       primedec -> idealprimedec
       racine -> sqrtint,
       regula -> quadregulator
       rename cmp_ZV -> ZV_cmp
       smith2 -> smithall, gsmith2 -> gsmithall
       srgcd -> RgX_gcd
       subfields0 -> nfsubfields,
       subres -> resultant, subresall -> resultant_all.
       torsell -> elltors and document it
       zideallog -> ideallog
       RgXQ_caract -> RgXQ_charpoly
       ZX_caract -> ZXQ_charpoly
       ZX_isirreducible -> ZX_is_irred
       pariputc -> pari_putc, pariputs -> pari_puts, pariflush -> pari_flush,
       gpmalloc -> pari_malloc, gprealloc -> pari_realloc, gpfree -> pari_free
       ordell -> ellordinate
    7- polvaluation -> RgX_valrem(x, pz) and no longer accept pz = NULL
       (use RgX_val). Same for ZX_valuation -> ZX_valrem [ & ZX_val ]
    8- rename u2toi -> uu32toi [ paste two 32-bit unsigned into a t_INT ]
       Write uutoi for the more natural operation of pasting two ulongs into
       a t_INT.
    9- prototype of FpX_center [ add one argument ]
BA 10- [svn/CVS] number development snapshots according to repository version
   11- delete CVS.txt and replace it with README-subversion
   12- syntax errors, SIGINT and "PARI bugs" are no longer trapped
   13- quadray no longer allows an (optional) 3rd argument
   14- 'format' was updated after each 'realprecision' change. Don't update
       if 'format' is Xm.-1 (show all significant digits).
   15- the algorithm used to round t_REAL for output
   16- the "minimum field width" component of the 'format' default is now
       ignored (was used only for integers in 'prettyprint' output mode and
       defaulted to 0 [no effect]). Use printf !
   17- %Z is no longer a valid conversion specification for PARI formats,
       since this is now handled as a length modifier. Use %Ps instead
       (P stands for PARI).
   18- remove the (ugly) prettyprint engine ('output' = 2), render as matbrute.
       printp / printp1 now act as print/print1
   19- pariprintf renamed to pari_printf, %Z conversion disappeared (use %Ps)
   20- check for small prime divisors in ispower [ large speedup ]
   21- the conversion style 'g' of the 'format' default now printfs in style
       'f' if the decimal exponent is < -4 ( was: if the binary exponent
       is < -32 ), in order to conform to standard printf specifications.
   22- [GMP kernel] enable GMP exact division
   23- matdetint: use less memory + faster finish
   24- change the default algorithm for mathnf(a) [ much faster ], flag 0 is
       no longer deprecated
   25- library/GP function galoisinit() now returns NULL/0 on failure
       (used to raise an exception)
   26- allow nfgaloisconj(t_POL, 1) [ cf C-4 ], never resort to nfgaloisconj(,2)
       [ slow, unreliable ] with default flag = 0
   27- [library]: gsub now implemented directly when typ(x) = typ(y)
BA 28- all LLL variants now use an implementation of NGuyen & Stehle's algorithm
   29- [library]: the "prec" parameter of floating point LLLs has disappeared
   30- Vecrev(x) now equivalent to extract(Vec(x), "-1..1") [was only for t_POL]
   31- allow t_COMPLEX as well as t_QUAD in ellpow
   32- we used to have Set(1) == Set("1") ( = ["1"] ). Now the latter returns
       ["\"1\""] (the Set whose element evaluate to the character string "1")
   33- plotinit: change the way default arguments are evaluated (omit, instead
       of interpreting 0 in a special way)
   34- nfbasistoalg / nfalgtobasis no longer accept t_VEC/t_COL/t_MAT,
       use matbasistoalg / matalgtobasis, which now also accept t_VEC/t_COL
   35- uniformize factornf and nffactor outputs (+ speed them up for
       non-squarefree inputs: compute valuations mod a suitable prime)
   36- treat Mat(t_LIST) as Mat(t_VEC)
   37- t_POL^0 always returned t_INT 1 [ try to return 1 in coefficient ring ]
   38- polresultant(x,y,,0) now chooses the best algorithm by itself
       [Ducos, modular, Sylvester], polresultant(x,y,,2) is now a synonym
       for the default. The change also improves poldisc().
   39- regression tests in src/test/in/ no longer need to end with
       'print("Total time spent: ",gettime)'
   40- arguments of stirling1 / stirling2 are now ulongs [ were longs ]
BA 41- user variables can now be aliased [#790]
   42- conjvec(POLMOD of INTMODs) was slow
   43- bnfcertify can now handle Zimmert/Minkowski bounds > 2^31
   44- moved misc/dico.new to src/whatnow + integrate it in the
       description system [ new script desc/whatnow ]
   45- bittest(x, n) no longer accepts t_VEC arguments n
   46- split library function disable_dbg() -> dbg_block() / dbg_release()
   47- make the semantics of substpol more precise and less surprising:
       substpol((x^2+x+1)/(x^2+x-1),x^2,x^3) -> (x^3 + x + 1)/(x^3 + x - 1)
       [ was: (x^6 - x^3 - 2*x - 1)/(x^6 - 3*x^3 + 1)  ]
   48- deprecated flag '2' for bnfinit [ don't compute units ] is now ignored
       ==> bnfunit becomes totally useless.
   49- the last component in prime ideals "primedec" format is a number field
       element, but no longer necessarily in t_COL form
   50- no longer allow ideals in *non-square* t_MAT form
   51- idealmin now returns a number field element, instead of the associated
       "principal idele"
   52- cleared up the confusion caused by PARI "ideles", which were remotely
       related to the mathematical notion, and covered two distinct uses. The
       first one (ideal + list of real places) is now called "divisor"; the
       second one (ideal + factorization matrix, which used to be a vector
       of logarithms of complex embeddings) is now called "extended ideal".
   53- Allow Scalar + t_COL: x + y returns [ y[1]+x, y[2],... ]. Consistent
       with RgC_Rg_add and RgV_isscalar, and allows linear algebra over number
       fields with less rigid data types (can mix freely t_COL, t_INT
       and t_FRAC, instead of converting everything to t_COL).
   54- Functions quadunit(), quadregulator(), and factor() no longer apply
       componentwise to vector / matrix arguments. Use apply()
   55- [libpari] removed obsolete 'prec' argument from prototypes of idealmul0,
       idealmulred, idealpow0, idealpowred, ideallllred, algdep, algdep0,
       lindep, lindep0
   56- factorback() no longer accepts an optional 3rd argument (nf). Use
       idealfactorback(). See also nffactorback().
   57- Extend polhensellift to work over unramified extensions of Q_p
       [ was Q_p only ]
   58- [libpari] changed prototypes of bnrdisc, bnrconductor, bnrisconductor
   59- slightly change the matrices used in idealmin (round them): the elements
       returned are a little different.
   60- no longer export private library function incgam2. Remove it also
       under GP when compatible = 3. Like incgam1 (suppressed), 1.39.15 doc
       stated it was provided "for debugging only".
   61- addhelp: TABs are now treated properly (would overflow line without
       wrapping) and \n are no longer ignored. No longer add a trailing '.'
       when missing.
   62- Useless nfinit flags 4, 5 (partial polred) removed.
   63- Split off inline functions unrelated to the kernel from
       src/kernel/level1.h
   64- Swap the order of the 2 arguments to select(): selection function
       now comes first [ as in apply() ]
   65- error() messages now prefixed with the customary *** [ used to be ### ]
   66- dvmdiiz(x, y, z) now expects z to be a t_INT [ to be consistent with
       all other xxxz functions where xxx includes only i,s,u arguments ]
   67- improved charpoly(,,2) [ Hessenberg form ]
   68- allow Flxq_mul and Flxq_sqr to call Flx_rem_montgomery
   69- t_POLMOD * t_POLMOD with same modulus: if all coefficients are in Z/NZ,
       call specialized functions FpXQ_mul / Flxq_mul [ analog with gsqr, gdiv,
       ginv, powgi ]
   70- bezout(): make sure the leading term of the gcd of 2 polynomials is positive
       (when it makes sense), as was already the case for gcd()
   71- mpcopy, Flv_copy, Flx_copy, vecsmall_copy have all become
       aliases for leafcopy.
   72- trap(,,foo) was actually equivalent to trap(,foo) changing the global
       exception handler, which was awkward an unintended
   73- norml2(t_POL) now returns the expected (square of) the standard L^2 norm.
       Use x * conj(x) to get back the old behaviour. norml2 now raises an error
       on t_POLMOD and t_FFELT components; used to add relative norms as in
        ? norml2([Mod(Mod(1,3)*x,x^2-2), 2])
        %1 = Mod(2, 3)
   74- [library] kernel functions involving a t_INT and a t_REAL now return a
       t_REAL, e.g. divsr(0, x) or mulir(gen_0,x) return real_0(...) [ used
       to return gen_0 ] ==> much better control of object types when writing
       kernel code. Generic functions (gmul, gdiv), as called from gp still
       return a result as precise as possible given the input: 0 * 1. --> gen_0
   75- [library] cleanup t_COMPLEX arithmetic to take advantage of real parts
       equal to a t_INT 0. Imaginary parts equal to 0 should never be created by
       our generic functions (so no need to cater for them specially: they
       will be very rare, produced by ad hoc routines).
       Special case trigonometric functions on pure imaginary arguments.
   76- kill(z) no longer destroys user variable, only resets them to their
       "undefined" value (as z = 'z would).
   77- [library] prototype of galoisconj [ add one argument ], and made it
       correspond to the default nfgaloisconj flag
   78- allow ordinary sorted vectors in set* functions [ sets are still t_VECs
       of t_STR, but if components can be directly compared, ordinary vectors
       are simpler and faster ]
   79- allow arbitrary nf entries in relative matrices; nfhnf and nfhnfmod
       return matrices whose entries may be t_INT or t_FRAC besides the
       traditionnal nfalgtobasis form. Allow ideals in arbitrary form [was:
       necessarily HNF]
   80- linear algebra routines: determines first whether we are over Z/NZ
       then call specialized FpM_* routines if possible [fixes #376]
   81- [library] commented out a large section of pariold.h. Define
       PARI_OLD_NAMES to recover compatibility macros for old friends like
       'un' and 'lstoi'
   82- Mat(t_QFI or t_QFR) now returns the associated 2x2 symmetric matrix
   83- [library] rename Buchall -> Buchall_param and export a new Buchall
       with a simplified interface.
   84- trap() no longer allows installing default error handlers.
   85- GP's break loop is now *enabled* by default [ set breakloop = 0 in your
       gprc to disable it ], and no longer controlled by trap(). To get out
       of a break loop, hit Return three times (next, return, break are no
       longer accepted)
BA 86- factormod and matker: faster linear algebra over F2.
   87- remove duplicate polynomials in polred output [#874]
   88- let forvec(X = v,...) respect the type of v (t_VEC or t_COL)
   89- polred: compute characteristic polynomials as in polredabs [ using
       complex embeddings, not ZX_charpoly ]
PMo90- algorithm for nfrootsof1 now defaults to nffactor: generically a little
       slower than qfminim, but safer and sometimes much faster (e.g. 1mn
       against a few hours)
   91- added a new component nf_get_roundG to the nf structure [ to speed up
       ideal reductions ]
BA 92- make test-xxx: trap SIGINT in benching script to exit on <C-C>
   93- removed optional flag to rnfconductor, use rnfisabelian.
   94- improved dirzetak() by an order of magnitude: e.g. K=nfinit(x^3-10*x+8),
       dirzetak(K, 10^6) from 4min (2.4.2) to 2.5s
   95- renormalize rnfkummer output so that it cannot be written as
       P(C * x)*C^(-deg P) for any integer C > 1  [ saves large powers of C
       in disc P ]
   96- added a flag to the (mostly useless) function rnfdedekind(), the
       version with the flag set is a little more useful. Allow list of prime
       ideals instead of single prime ideal, and make this list optional
       (omitted = test for maximality at ALL primes)
   97- the "zetakinit" format: bnf no longer included
   98- thueinit / thue: allow reducible and non-monic polynomials
   99- rnfdedekind: allow non-monic polynomials
  100- bnrclassno(bnr,id) : allow bid as well as modules for the 'id' argument.
  101- GP 'path' now also applies for file paths containing a '/' provided
       they do not *start* with a '/' (i.e. not an absolute path)
  102- [library mode] the function znlog now takes an extra "order" argument
  103- added an optional "order" argument to znlog, fflog, elllog, znorder,
       fforder, ellorder, whose preferred format is [ord, factor(ord)] and
       ord is the order of the group
  104- factor(t_INT/t_FRAC, lim) used to trial divide by primes up to
       min(lim, primelimit). Now we trial divide up to lim and raise an error
       if lim > primelimit. [ Having the routine return an obviously "wrong"
       result depending on an invisible parameter was not a good idea. ]
  105- factor(t_INT/t_FRAC, 1) was the same as factor(t_INT/t_FRAC, 0). Now
       abide by the input value and leave (..., 0) as the single special case
       (shortcut for "the largest precomputed prime")
BA106- [native kernel] Schoenhage-Strassen multiplication/squaring
  107- factorff(A, p, T): make p & T optional if A has t_FFELT coefficients
BA108- [TLS] Change pari_thread calling convention
  109- allow x.a1, ..., x.a6 for an "ell5" x (5-component vector)
  110- removed 'primelimit' grom gp_data struct: we now print the actual limit
       of the primetable, not the value input by the user (which may be a
       little less)
BA111- listsort(L) no longer returns the list L: it now returns nothing.
       (No point in sorting in place if we must immediately copy the result.)
  112- allow component(t_VECSMALL,...) [#957]
  113- allow bestappr(x)  [2nd argument made optional]
BA114- GMP kernel chosen by default when GMP is available
BA115- Rename RgXQ_u_pow to RgXQ_powu
  116- Darwin: add -search_paths_first to LDFLAGS (try to solve the Editline
       / Readline conflict in a user-friendly way)
BA117- [enable-tls] Library soname changed to libpari[-gmp]-tls[-2.4].so.N.
  118- integrate a final polredabs in quadhilbertreal()
BA119- Rename leftright_pow to gen_pow, leftright_pow_u to gen_pow_u and
       implement sliding window powering.
BA120- galoisapply now accept automorphisms in t_COL format.
  121- allow moduli equal to 0 in matsolvemod (was SEGV) [#947] + rigorously
       size-reduce the output in all cases.
  122- the bnr structure in a minor way (bnr[6] has 3 components now)
  123- cleaned up the code used by the GP evaluator to call PARI (and
       installed) functions. Functions can now have 20 arguments [#795]
BA124- galoisfixedfield now accept subgroups.
  125- ellap(E, p) second argument now optional if E has t_INTMOD or t_PADIC
       coefficients.
  126- generalize the "generic pivot" strategy to all linear algebra (support
       RgM_solve and det() in addition to existing functions); implement
       maximal pivot strategy for p-adics [#1054]
BA127- padic_sqrt -> Qp_sqrt, padic_sqrtn -> Qp_sqrtn, gammap -> Qp_gamma
  128- rename rnfinitalg -> rnfinit
  129- [readline + vi-mode] h/H no longer clobbered by online help. 'h' will
       now act as expected (= 'move left one character')
  130- any of break / next / return now get out of the break loop debugger
       (as they used to in 2.3)
  131- typing <Return> thrice no longer gets out of the break loop : use break
  132- faster eta(,1), weber() and quadhilbert() [ rewrite using eta's
       functional equation explicitely ]
  133- checked that no 2-strong pseudoprime passes BSPW primality test up to
       2^64 (use Feitsma's list, extending Galway's)

  Added
    1- many new benches (make test-all)
    2- split up some large monolithic files: new files
       src/basemath/ RgV.c, FpV.c, ZX.c, ZV.c, concat.c, lll.c, hnf_snf.c
         prime.c, bit.c
       src/modules/ QX_factor.c, Hensel.c, DedekZeta.c
    3- split off module of general inline functions: src/headers/pariinl.h.
    4- Write or rename / document many convenience library functions:
       pari_calloc, cgetalloc
       mkfraccopy, mkquad, mkvecsmall4, absfrac, sqrfrac, mul_content, mul_denom
       roundr_safe, trunc_safe
       shallowextract, shallowconcat1, concat1,
       BPSW_psp & BPSW_isprime
       Z_smoothen, Z_ispower, Z_ispowerall
       Z_isanypower, chinese_coprime_Z, Z_chinese, Z_chinese_all,
       Z_chinese_coprime, Z_chinese_post, Z_chinese_pre, Z_incremental_CRT,
       Z_init_CRT, ZM_incremental_CRT, ZM_init_CRT, ZX_incremental_CRT,
       ZX_init_CRT,
       ZX_to_monic, ZX_primitive_to_monic, ZX_squff,
       ZX_gcd_all, nfgcd, nfgcd_all,
       RgX_disc, quad_disc, qfb_disc,
       ZpX_liftfact, ZpX_gcd, ZpX_reduced_resultant, ZpX_reduced_resultant_fast
       Fp_center, FpM_center, FpV_center
       Fp_ratlift, FpM_ratlift, FpX_ratlift
       FpM_FpC_mul_FpX
       Fp_red, FpXQ_red
       Flxq_add, Flxq_sub, FpXQ_add, FpXQ_sub, FqXQ_add, FqXQ_sub
       FqX_add, FqX_sub
       FqX_roots, FpX_factorff, FpX_rootsff
       FpXQX_div, FpXQX_rem, FpXQXQ_mul, FpXQXQ_sqr, FpXQXQ_pow
       FpXQXQ_inv, FpXQXQ_invsafe, FpXQXQ_div
       FqXQ_mul, FqXQ_sqr, FqXQ_pow, FqXQ_inv, FqXQ_invsafe, FqXQ_div
       FqX_extgcd
       ZC_ZV_mul, ZM_is_identity, ZM_copy, ZM_neg, ZM_add, ZM_mul, ZM_pow
       ZM_max_lg, ZM_sub, ZM_ZC_mul, ZXV_Z_mul, ZM_Z_mul, ZV_Z_mul, ZV_isscalar,
       ZV_copy, ZV_neg, ZV_neg_inplace ZV_togglesign, ZV_indexsort
       ZMrow_ZC_mul, ZV_ZM_mul, ZV_cmp0, ZV_content, ZV_equal,
       ZC_lincomb, ZC_lincomb1_inplace, ZC_z_mul,
       ZV_prod, ZV_pval, ZV_pvalrem zv_prod
       ZM_Z_divexact, ZC_Z_divexact, ZX_Z_divexact,
       ZM_charpoly, Flm_charpoly, Flm_hess
       Z_ZX_sub, ZX_Z_sub, Fp_FpX_sub,
       Flc_Fl_div, Flc_Fl_div_inplace, Flc_Fl_mul, Flc_Fl_mul_inplace,
       Flm_Fl_mul, Flm_Fl_mul_inplace, Flm_copy, Flm_mul, Flm_inage,
       Flv_Fl_mul_inplace, Flv_add, Flv_add_inplace, Flv_copy, Flv_dotproduct,
       Flv_sub, Flv_sub_inplace,
       Fl_order, Flx_nbfact_by_degree, Flx_roots_naive,
       FF_sub,
       Fp_mulu,
       FpV_inv, FpXQ_inv, FqV_inv,
       FpXY_eval, FpXY_evalx, FpXY_evaly,
       Flx_Fl_mul_to_monic, FlxqX_Flxq_mul_to_monic, FlxqX_rem,
       FlxqX_gcd, FlxqX_extgcd,
       FlxqXQ_mul, FlxqXQ_sqr, FlxqXQ_inv, FlxqXQ_invsafe,
       pol1_FlxX,
       FpX_Fp_sub, FpX_Fp_sub_shallow,
       FpX_Fp_mul_to_monic, FqX_Fq_mul_to_monic,
       FpXX_Fp_mul,
       RgM_check_ZM, RgM_det_trianguar, RgM_isdiagonal,
       RgV_check_ZV, RgV_neg, RgV_add, RgV_sub, RgV_Rg_mul,
       RgC_neg, RgC_add, RgC_sub, RgC_Rg_div, RgC_Rg_mul,
       RgM_neg, RgM_add, RgM_sub, RgM_Rg_div, RgM_Rg_mul,
       RgC_RgM_mul, RgC_RgV_mul, RgM_RgC_mul, RgM_RgV_mul, RgM_mul, RgM_sqr,
       RgV_RgM_mul, RgV_RgC_mul, RgV_add, RgV_neg, RgV_sub, RgC_Rg_add
       RgM_diagonal, RgM_diagonal_shallow, RgM_inv, RgM_solve,
       RgM_mulreal, mulreal, RgX_RgM_eval
       RgV_sum, RgV_sumpart, RgV_sumpart2
       RgX_neg, RgX_add, RgX_add_shallow, RgX_sub, RgX_Rg_add, RgX_Rg_sub,
       Rg_RgX_sub, RgX_equal, RgX_equal_var, RgX_translate,
       RgX_to_nfX, RgM_to_nfM, RgC_to_nfC,
       RgX_type, RgX_type_decode, RgX_type_is_composite, RgX_Rg_divexact,
       RgXQ_norm, RgXQ_pow, RgXQ_caract, RgXQ_reverse
       RgXQX_translate, RgXQV_to_mod, RgXQXV_to_mod
       RgXQX_pseudorem, RgXQX_pseudodivrem,
       RgX_check_ZX, RgX_check_ZXY, RgX_pseudodivrem, RgX_pseudorem,
       RgX_recip, RgX_recip_shallow,
       ZC_hnfrem, ZC_hnfremdiv, ZM_det_triangular, ZM_detmult, ZM_equal, ZM_hnf,
       ZM_hnfall, ZM_hnfcenter, ZM_hnflll, ZM_hnfmod, ZM_hnfmodid,
       ZM_hnfmodall, ZM_hnfrem, ZM_hnfremdiv, ZM_hnfperm, ZM_snf, ZM_snfall,
       ZM_snfall_i, ZM_snf_group, ZM_snfclean
       ZM_lll, ZM_lll_norms, hnfmerge_get_1,
       ZXY_max_lg, ZX_copy, ZX_equal, ZX_max_lg, row_Flm, row_zm,
       scalar_ZX, scalar_ZX_shallow, deg1pol, deg1pol_shallow
       togglesign_safe, setabssign,
       zero_Flm, zero_Flv, zero_zm, zero_zv, zm_copy, zv_cmp0, zv_copy,
       zv_equal, zv_content, zx_renormalize,
       identity_perm, cyclic_perm, perm_mul, perm_commute, perm_inv, perm_pow,
       cyc_pow_perm, cyc_pow, perm_cycles, perm_order, vecperm_orbits
       bitvec_test_set,
       RgX_get_0, RgX_get_1,
       resultant, resultant_all, QX_disc, QX_resultant,
       lllfp, lllintpartial, lllintpartial_inplace, reducemodlll,
       reducemodinvertible, closemodinvertible,
       init_primepointer,
       galoisinit0, get_nfpol, get_bnfpol, get_prid,
       bnf_get_nf, bnr_get_bnf, bnr_get_bid, bnr_get_mod, bnr_get_nf
       nf_get_r1, nf_get_r2, nf_get_roots, nf_get_sign, nf_get_roots,
       nf_get_M, nf_get_G, nf_get_roundG,
       nf_get_Tr, nf_get_TrInv, nf_get_disc, nf_get_index, nf_get_pol, nf_get_zk
       nf_get_prec (was nfgetprec), nf_to_scalar_or_basis, nf_to_scalar_or_alg
       nf_to_Fq_init, nf_to_Fq, Fq_to_nf, zkmodprinit, zk_to_Fq_init, zk_to_Fq,
       nfM_to_FqM, FqM_to_nfM, nfX_to_FqX, FqX_to_nfX
       zk_multable, zk_scalar_or_multable, multable, tablemul, tablemul_ei,
       tablemul_ei_ej, tablemulvec, tablesqr, ei_multable.
       nfordmax, nfarchstar (was zarchstar), nfadd, nfsign, nfnorm, nftrace
       nfsign_arch, nfsign_units, nfsign_from_logarch, Buchquad, Buchall,
       Idealstar, idealaddtoone_i, idealcoprimefact,
       nfinvmodideal, nfpowmodideal, idealsqr, nfpow_u
       numberofconjugates,
       bnrisconductor0, bnrdisc0, bnrconductor0,
       bnrsurjection, ABC_to_bnr, idealred, idealred0 (was ideallllred),
       idealred_elt, idealred_elt0
       ellinit_padic, ellinit_real, ellinf, ell_is_inf, ell_is_padic,
       ell_is_real, checkerr_real, checkell_padic, checksmallell, checkell5,
       checkellpt
       trueeta, exp_Ir, roots_from_deg1, deg1_from_roots
    5- [libpari] public generic hashtables
    6- library functions sqrr and invr [ Newton inversion, called by ginv() ]
BA  7- library functions checkgroup, galois_group, cyclicgroup, dicyclicgroup,
       abelian_group, group_domain, group_elts, group_order, group_isabelian,
       group_abelianHNF, group_abelianSNF, group_isA4S4, group_leftcoset,
       group_rightcoset, group_perm_normalize, group_quotient, quotient_perm,
       quotient_group, quotient_subgroup_lift, group_subgroups,
       groupelts_abelian_group, groupelts_center, group_export, group_ident
    8- library functions pr_get_p, pr_get_gen, pr_get_e, pr_get_f, pr_get_tau
       to access components of prime ideals
BA  9- handle numerical derivation as an operation on closure.
RB 10- GP functions printf, Strprintf
   11- pari_printf, pari_fprintf, pari_sprintf, pari_vfprintf, pari_vprintf,
       pari_vsprintf
BA 12- docpdf and install-docpdf targets (toplevel Makefile)
JD 13- add an optional argument to quit()
   14- universal constant gen_m2, for symmetry
   15- GP function version() [ current version number ]
BA 16- [ellap] Port of SEA algorithm from the ellsea GP package
BA 17- new alarm GP function and alarmer error code
RM 18- partitions()
   19- New chapters in Libpari Guide: usersch6.tex, usersch7.tex
   20- library function dbg_pari_heap
   21- function bnfcompress() [ to create a "small bnf" from a true
       bnf, to be recoved using bnfinit ]. More consistent than old
       bnfnit(,3) / bnfmake.
   22- a concept of "extended ideal" [I, t], where t is an algebraic number
       (possibly in factored form). The pair represents the ideal I x (t)
       In applications the norm of I stays bounded for a given base field,
       and t keeps track of the "principal part" of the ideal (usually in
       factored form to avoid coefficient explosion).
   23- public interface to forsubgroup: traversesubgroups()
   24- library function bnfisprincipal0 (with new public flags, see nf_GENMAT)
   25- GP functions nffactorback / idealfactorback
JD 26- GP function warning()
   27- new flag to charpoly [ integral matrix, modular algorithm ]
BA 28- GP function galoisisnormal()
BA 29- F2x functions family for polynomials over GF(2) and
BA 30- t_FFELT subtype_FF_F2xq for GF(2^n)
BA 31- F2v/F2m functions family and F2m_ker for linear algebra over GF(2)
BA 32- extra debugging data in t_CLOSURE and debugging facility.
BA 33- compile-time copy optimizer
BA 34- elliptic discrete logarithm function elllog()
BA 35- direct implementation of ZX_mul, ZX_sqr using Kronecker's trick + mulii
   36- rnfisabelian GP function
   37- default 'recover'
   38- more possibilities to draw "random" elements in natural sets using
       random(), e.g. finite fields, elliptic curve over finite field.
   39- GP routines nfeltadd, nfelttrace, nfeltnorm
BA 40- Pollard rho discrete logarithm algorithm (function gen_Pollard_log)
   41- GP routine polrootsff
   42- matadjoint: implement division-free algorithm (flag = 1) [#937]
BA 43- GP routines and PARI functions ellweilpairing, elltatepairing and
       ellgroup
BA 44- FpE functions family for points on E(F_p)
BA 45- GP routine galoisgetpol and PARI function galoisgetpol, galoisnbpol
BA 46- GP routine and PARI function elldivpol
BA 47- GP routine and PARI function ellmodulareqn
BA 48- function RgXQ_ratlift for reconstruction of rational functions.
   49- optional flag to bnfcertify : only check that the correct class group
       is a quotient of the computed one.
VB 50- mingw support
VB 51- plotwin32 graphic engine
   52- plotQt4 graphic engine
BA 53- low-level accessors functions gal_get_* for galoisinit objects
   54- libpari functions cvstop2, z_lval, z_lvalrem, z_pvalrem
   55- GP function externstr()
   56- library function ZC_reducemodmatrix, ZM_reducemodmatrix,
       ZC_reducemodlll, ZM_reducemodlll
BA 57- libpari functions vecsmall_max and vecsmall_min
BA 58- GP functions idealfrobenius and idealramgroups
BA 59- library functions Zn_sqrt, Zn_issquare
   60- library function gidentical
BA 61- GP operator === that use gidentical
   62- GP function sumdedekind()
JD 63- [gphelp] support for bzip2 compressed documentation.

  Removed
    1- obsolete GP functions printp, printp1. Either replace them by
       print / print1 [ or the new printf ], or define aliases
       alias(printp, print); alias(printp1, print1)
    2- obsolete GP functions bnfclassunit, bnfreg, bnfclgp, bnfunit.
       Use bnfinit.
    3- obsolete GP functions bnrclass. Use bnrinit or bnrclassno.
    4- obsolete GP function idealprincipal. Use number field elements "as is"
    5- obsolete GP function ideleprincipal. No routine remains that would
       use the auxiliary "archimedean information"
    6- obsolete library functions extract and matextract
    7- obsolete optional argument to quadhilbert(D < 0)
    8- unused undocumented library functions os_open/os_read/os_close
    9- obsolete README.DOS and README.WIN
   10- cant_deflate, elliper1, lllger3, varer1, obsoler, infprecer, errlg,
       errexpo, errvalp, rtodber, affer2, primer2, siginter error codes
   11- obsolete library function bruteall
   12- useless undocumented library function gkrogs (use krois)
   13- obsolete library functions roots2 and rootsold are no longer public
   14- useless library function isnfscalar
   15- obsolete undocumented library function factpol(), lift_to_pol()
   16- library function combine_factors() is no longer public
   17- obsolete undocumented library function rnfdet0, rnfdet2
   18- obsolete library function discsr(). Use RgX_disc (or quad_disc, qfb_disc)
   19- obsolete library function element_powmodidele() [use element_powmodideal
       + set_sign_mod_idele], ideleaddone, reducemodmatrix, nfreducemodidele,
       nfreducemodideal, nfreducemodideal_i, elementinv_modpr
   20- useless macros gaddgsz, gaddsgz, gdiventgsz, gdiventsgz, gdivgsz,
       gdivsgz, gmaxgsz, gmaxsgz, gmingsz, gminsgz, gmodgsz, gmodsgz, gmulgsz,
       gmulsgz, gsubgsz, gsubsgz, gachz, gacosz, gashz, gasinz, gatanz, gathz,
       gchz, gcosz, gcotanz, gexpz, ggamdz, ggammaz, glngammaz, glogz, gpsiz,
       gshz, gsinz, gsqrtz, gtanz, gthz, gzetaz, gmaxz, gminz, TRgopgz,
       mpfloorz, mptruncz, mpdvmdz
   21- obsolete library function gmillerrabin, gpolcomp, caradj0
   22- obsolete library functions allbase, base, base2, factoredbase, smallbase,
       discf2, factoreddiscf, smalldiscf. Use nfbasis0 / nfdisc0 / nfmaxord
   23- obsolete library function pari_rand31  [ use pari_rand ]
   24- obsolete undocumented library function bruterr
   25- obsolete library function smallbuchinit
   26- awkward flag '3' for bnfinit. Use bnfcompress().
   27- badly named bnfmake. Use bnfinit(sbnf).
   28- useless component v[6] in "small bnfs" (was 'different', is 0)
   29- obsolete library functions buchray, buchrayinit, buchrayinitgen:
       use Buchray.
   30- obsolete library function smallfact [ use boundfact or Z_factor_limit ]
   31- obsolete library function gracine [ use sqrtint ]
   32- obsolete library function ispsp, gispsp. Use ispseudoprime or BPSW_psp.
   33- obsolete library function gregula, gfundunit, gboundfact.
       Use quadregulator, quadunit, boundfact
   34- obsolete library functions ideallistunit, ideallistunitgen,
       ideallistzstar, ideallistzstargen. Use ideallist0.
   35- obsolete library function algdep2. Use algdep0 (same arguments).
   36- obsolete library function Mod0. Use gmodulo.
   37- obsolete undocumented library function poldvd. Use
       RgX_divrem(x,y, ONLY_DIVIDES)
   38- obsolete library function kerint1. Use kerint or ZM_lll(,0.99,LLL_KER)
   39- header file paritype.h [ split between parigen.h and pariinl.h ]
   40- obsolete library functions sor, outbrute, outbeaut
   41- dangerous macros max & min [ use maxss/minss, maxuu/minuu, maxdd/mindd ]
   42- deprecated global constants gi, geuler, gpi. Use
       mulcxI/mkcomplex/gen_I(), mpeuler() and mppi().
BA 43- bitvec family function. Use F2v instead.
BA 44- useless prototype code 'S'. Use 'r' instead.
BA 45- useless function delete_named_var.
   46- obsolete and dangerous switch_stack() [#1013]

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Done for version 2.4.2 (released 22/12/2007):

  Fixed
    1- divrs(0., negative integer) had the wrong exponent
    2- vecsort(,,4) was broken in 2.4.1
    3- allocatemem(z) didn't check that z >= 0 [#556]
    4- remove hack in internal function 'readbin' that used the clone bit as
    a special marker hence returned an invalid object, which could be
    propagated to higher level public functions.
BA  5- sqrtn(Mod(a,p),..) and factorff() were broken in 2.4.1
BA  6- bitnegimply(1,2^65) returned a wrong result [#560]
    7- nfeltreduce(nf,t_POLMOD,id) didn't work [#558]
    8- [library] missing function intfourierexp()
    9- segv in polrootspadic / repeated roots in FpX_root (only in deg 2) [#562]
   10- log(2+O(4^4)) --> oo loop
   11- log(exp(1e-100)) -> 1.000000000149525401279188592 E-100
       [ for some accuracies, log & exp inacurate near 1 and 0 resp. ]
   12- numbpart inaccurate (and slow). E.g numbpart(52602) off by 1 [#557].
BA 13- &, |, && and || were right-associative instead of left-associative
   14- remove type assumptions in mulcxI, mulcxmI: unsafe [#516]]
   15- too much memory allocated to print real numbers
   16- (y/x)/(x^0/y) not correctly simplified
   17- \s reported a misleading number of available user variables
   18- 1/(quadgen(-4)*x)+0. -> exponent overflow [#591]
   19- sumpos(x=1,1/x^2,1) had the wrong sign [#587]
   20- matrixqz(x,0) would not work if first two determinants were 0
       + might lose pointers in gerepile
   21- sumpos(x=1,0) -> oo loop [#585]
   22- qfminim([[;]],,,2) --> SEGV [#598]
   23- intformal(1) --> incorrect object
   24- intformal(y,x) --> y*y
   25- matadjoint(0) --> gerepile error
   26- prodinf(x=0,0), prodinf(x=0,-1,1) --> oo loop
   27- agm(-1, 1+1e-55) --> oo loop, agm(-1,1) --> wrong result
LGr28- killing hi-res graphic window created zombie process (missing wait())
   29- missing type check in eint1(x,n)
   30- ploth(...) + Ctrl-C --> hi-res graphic window killed [use pari_daemon()]
   31- quadgen([]) --> incorrect object [#606]
   32- hyperu(0,1,1) --> oo loop [#608]
   33- fix headers so that ulong is always defined
   34- read("a b.gz")   \\ filename contains a space
       gzip: ./a.gz: No such file or directory
   35- logagmcx inaccurate [ used for log(t_COMPLEX), large precision ]
   36- RgX_divrem: prevent "impossible" situations for weird base rings
       1/ Mod((y-1)*x-1, Mod(2,6)*x^3-Mod(1,2)*x^2-1) --> SEGV
   37- gclone didn't work on t_INTs with lg > lgefint
   38- GC error in add_rfrac() [#612]
   39- missing type check in subgrouplist [#616]
   40- inconsistent type checking in nfeltdivmodpr [#617]
   41- nfdisc(x^32+24888960*x^30+227077545803904*x^28+887225199431341086720*x^26+1145458263702669503266741248*x^24-459489127319699704284694158213120*x^22+73430099675988831740428200872826863616*x^20-6599321778961370851005469933592282336329728*x^18+369563540304176984501448638025309170375722401792*x^16-13214618078037183940422584396181416089308059714715648*x^14+291247424536170563676138246427605494527806096141868597248*x^12-3407227562250271661213064368167141844394234574629997803208704*x^10+8699659624155196549893192408316795253695601521375537680749690880*x^8+153390110447793800411208197070089975539253907830958400507847198638080*x^6-34704742472413552993642506624568826441560550586777137133227202205188096*x^4+43389826962123095743039577910848855441856968086933919852164410699581227008*x^2+336462877895453750216450425189196151877685043455706101021513167474262016, 1)
       --> "not a prime" error (help Round4 to recover when using non-primes,
       instead of raising immediate errors) [#624]
   42- vecextract: C14 in 2.4.0 didn't work for large masks [#623]
   43- clean up version handling: move version setting code to config/version
       and always set PARIVERSION from CHANGES if 'CVS' [ used to require
       a 'Configure' ]
   44- zeta(0e1) --> SEGV [#627], exp(0e1*I) --> SEGV [#630]
   45- exp(2^200*I) --> catastrophic cancellation [cos/sin were OK] [#631]
DE 46- on NFS filesystems, make install would rebuild all [#503]
   47- default(compatible,3); default(compatible,1) --> case no longer taken
       into account [#629]
   48- missing GC in 'sigma'
   49- eta(x): valuation would overflow if seriesprecision is large
   50- typo in src/kernel/none/mp.c:convi --> inefficiency
   51- concat(v) used too much memory (quadratic in #v, make it linear) [#634]
BA 52- gp -q -f < eval('y) --> SEGV
   53- "significant pointers lost" for objects involving 0 t_SERs [#635]
   54- for trivial x, isanypower(x,&p) would not set p
MSo55- [Configure] don't rely on $CC exit status, check whether a non-0 size
       executable is produced [ problems with tcc ]
BA 56- wrong Prototype for subgrouplist [ extra 'p' ]
   57- hole in 'secure' mode: don't allow changing psfile / logfile [#645]
   58- nf.codiff was only correct up to multiplication by some rational number
       [cf #510 & 2.4.1-F7]
   59- cgiv failed to delete the last object on stack [ if recursive ]
LGr60- first default(parisize,*) would segfault [due to C-1, #569]
   61- pariputs("") --> invalid read of size 1 [ valgrind ]
   62- ell.omega, ell.eta, pr.gen could lead to gerepile error [#641]
   63- y=[x];eval('y) --> error [#652]
   64- incorrect use of gerepileupto in inittestlift [ galoisconj ]
   65- extra multiplication in the innermost loop of Cholesky decomposition
       (qfgaussred, etc).
   66- made FpX_add, FpX_sub, Fq_add, Fq_sub stack-clean. FpX_neg, Fq_neg
   returned an incorrect result if input was not reduced mod p
   67- setrand(4);polgalois(x^8-264*x^6+25410*x^4-1054152*x^2+15856203)
       --> wrong result [#660]
   68- build fails with "env: parameter list too long" [#661]
   69- factorff did not accept inputs with t_FFELT coeffs
   70- could use PARI stack while reading gprc [ before pari_init ]
   71- Mod(Mod(1,v),v) --> invalid object.
   72- a = Mod(y, v); y = Mod(1,v); eval(a) --> invalid object
   73- for some complicated t_RFRAC z: z' --> impossible assignment I-->S
   74- typo in bnfisintnorm(): missed some solutions  [ couldn't find a unit
       of norm -1 even though one exists ]
   75- ffprimroot(ffgen( t_POL of degree 1)) --> oo loop
   76- wrong result in theta(q, z) if sin(nz) was small for some small n [#667]
   77- 1/Mod(0,2) --> impossible inverse modulo: Mod(2, 2)
   78- alias(a,b), then ?a --> 'a is aliased to a'
   79- -Mod(0,2) --> Mod(2,2)
MA 80- [Linux-PPC] missing -fPIC in DLCFLAGS
   81- possible oo loop in _isprincipal [ precision was supposed to increase,
       but could in fact remain the same ]
   82- quadregulator(y) -> SEGV [#674]
   83- acos(x^0) -> division by 0 [ instead of O(x^8), at \ps 16 ]
       Analogously, acosh(x^0), asin(x^0), asinh(I+O(x)) --> division by 0
   84- dilog(-1) [ more generally polylog of < 0 t_REALs ] should have
       0 imaginary part
   85- problems with  [ build/install ] directory names containing spaces
   86- avoid catastrophic cancellation in 3M [ Karatsuba ] formula for
       t_COMPLEX * t_COMPLEX
   87- ix86, x86_64: missing earlyclobber constraint for addllx, subllx, divll
   88- ploth(,4) --> huge memory use for large plots
   89- stirling(0,0,2) --> 0  [ should be 1, #690 ]
   90- deriv(x/(x+1),y) --> invalid t_RFRAC with exact 0 numerator [#687]
   91- issquare(t_POL) assumed characteristic 0 [#549]
   92- sqrt(Mod(4,5) + x) --> error [ e.g. stack overflow ]
   93- hyperu(0*I,1,1) --> forbidden assignment t_COMPLEX --> t_REAL.
BA 94- fix compilation problem with g++-4.2 and GMP.
   95- ??factor_proven, ??factor_add_primes did not work
   96- typo in znprimroot: wrong result for large moduli
       znprimroot(5*2^127+1) --> 2 [#696]
   97- ffgen(x*Mod(1,2)) --> x [ should be 0 ]
   98- ffprimroot(ffgen((x+1)*Mod(1,2))) --> oo loop
   99- nffactor(nfinit(y),x^2+Mod(1,y)) --> SEGV [#699]
  100- "precision error in minim0" on qfminim(G, norml2(M[,1]), 100)
       from tutorial
  101- nffactor(nfinit(y^2+1),(2*x+1)*x*(x+1)) --> SEGV [#702]
  102- isprime((6^2176+1)/(6^128+1)) --> length (lg) overflow [#697]
       [ analogous problems for any large integer ]
  103- various problems related to allocatemem() [ + document quirks ]
       Remove the br_ALOCATEMEM construct and end allocatemem0() by a longjmp
  104- missing GC in det_simple_gauss() [ matdet for inexact inputs ]
  105- rare stack corruption in add_rfrac [#700]
  106- add missing GC in gsubst
  107- polred([pol, b]) computed unnecessary but possibly expensive
       invariants of pol [ e.g. disc, index ]
  108- compilation failed on AIX [ YYSIZE_T, pow ]
  109- ? Mod(1,2)*x + Mod(1,2)*y
       %1 = Mod(1, 2)*x + (Mod(1, 2)*y) \\ extra parentheses in constant term
       ? % + 0*z
       %2 = Mod(1, 2)*x + Mod(1, 2)*y
  110- factornf(x^5+(-a+2)*x^4-a*x^3+(3*a-6)*x^2+(5*a-10)*x+(2*a-5), a^2-5)
       --> SEGV  [ not squarefree -> denominators creeping in ] [#708]
  111- problems with isexactzero and t_INTMODs. Mod(0,2)*x*1. -> 0,
       Pol(Mod(0,2))+2 -> 2, (2+0*I)+I*Mod(0,4)->Mod(2,4), Mod(0,2)/x -> 0
       Use isrationalzero instead.
  112- substvec(x^2, [x^2], [y]) --> y^2  [ should be an error ]
  113- typo in FpM_gauss_pivot: FpM_rank, FpM_image, FpM_suppl, FpM_indexrank
       much slower than they should be. Analogous problem in FqM_gauss_pivot.
BA114- missing GP2C descriptions for Pol and Polrev.
BA115- zero FFELTs were not considered as exact zeros [#710]
  116- rare SEGV in gp when recovering from error (dereferencing
       global_err_data equal to BREAK_LOOP)
  117- vecsort(t_LIST) returned a t_VEC
  118- gp "huge file" + stack overflow --> stack overflow in next
       interactive command (+ minor memoryleak) [#712]

  Changed
BA  1- The combined GP parser/evaluator has been replaced by a bytecode
       compiler and a bytecode evaluator
BA  2- install(): parser code 'E' and 'I' now refer to closures, not strings:
       'I': closure whose value is ignored, like in for() loop
       'E': closure whose value is used, like in sum() loop
    3- Fl_pow renamed to Fl_powu [ exponent may not be negative ]
    4- split usersch3.tex moving function documentation to src/functions/*
    5- simplify table of contents for users.dvi
    6- rename Flx_rand -> random_Flx, FpX_rand -> random_FpX
BA  7- use factor_pn_1 to compute various orders in FF.c
    8- file usersch3.tex is now generated from functions/*
    9- rewrote logr_abs [ inaccurate + slow for x = 2^k * (1 - epsilon) ]
   10- rewrote exp1r_abs [ inaccurate + slow ]  and mpexp_basecase [ reduce
       input mod log(2) ]
   11- rewrote mpsc1 [ slow ] -> faster sin and cos
   12- [library] rename pointch ->  ellchangepoint, coordch -> ellchangecurve
   13- prototype of constpi, consteuler.
   14- use a little less memory to compute pi, euler, log(2)
   15- qfminim(x,b,m,flag) made 'm' parameter optionnal also when flag=0
   16- made second argument to matrixqz optional, allow non-prime values
   17- matpascal(n < -1), mathilbert(n < 0) now raise an error
   18- add optional extra argument to ffprimroot, fforder, fflog
   19- allow znlog(x,g) where g a t_PADIC or an t_INTMOD modulo any N
       that znprimroot(N) would accept
   20- log(x t_PADIC): check whether x = 1 (mod p) before replacing x <- x^(p-1)
   21- znprimroot(p^k): use isanypower() instead of factor().
       E.g. znprimroot(nextprime(10^20)^1000): 8mn --> 12ms
       znprimroot(N) no longer checks reliably whether (Z/N)^* is cyclic.
       Result undefined if it is not.
   22- padic sqrt much faster [ small and large accuracies ]
   23- let primes() indicate a value of primelimit if unable to answer [#625]
   24- remove variable names from ?0
   25- exp(0e10) returned 1.000, made it 0exxx [ no significant digit ]
MSo26- define polchebyshev(n, 1 or 2), pollegendre(n) for n < 0
   27- znorder faster for non-prime modulus [ try it for Mod(2,5^10*7^10) ]
       (compute lcm of local p-adic orders)
   28- changed icopyifstack / copyifstack macros so that their arguments
       are GENs, not GENs typecast to long.
   29- add -funroll-loops to gcc flags when compiling arithmetic kernel
   30- improve ellap (Shanks) by computing #E(F_p) mod 2 [ idea stolen from
       Pierrick Gaudry ]
   31- nfreducemodpr was exceedingly slow for large t_POL inputs, e.g.
        w=x^48 + 158*x^46 + 4*x^45 + 12103*x^44 + 448*x^43 + 597874*x^42 + 23928*x^41 + 21373779*x^40 + 802424*x^39 + 588314524*x^38 + 18516794*x^37 + 12951694530*x^36 + 294992428*x^35 + 233870773964*x^34 + 2752210590*x^33 + 3524535272389*x^32 - 5797649292*x^31 + 44873186922754*x^30 - 798816466566*x^29 + 486736157075707*x^28 - 18082470992066*x^27 + 4523171646555185*x^26 - 271968456240780*x^25 + 36127625049532658*x^24 - 3144283847234232*x^23 + 248308835345289047*x^22 - 29271322082172250*x^21 + 1467438460133718165*x^20 - 223406933340907742*x^19 + 7435650911902583447*x^18 - 1405525828628464338*x^17 + 32139937168429173010*x^16 - 7272631544138987758*x^15 + 117563850386577478340*x^14 - 30686320483051428956*x^13 + 359649878031434743177*x^12 - 103986861631984042496*x^11 + 904392935429691313557*x^10 - 276283466933590000414*x^9 + 1822091999374665372405*x^8 - 554259164403897051340*x^7 + 2827767569115449802250*x^6 - 788757355446858093774*x^5 + 3170224521565069297504*x^4 - 708295322502763110380*x^3 + 2280480720610858280676*x^2 - 300602103002034938488*x + 788466592242441477569;
        nf=nfinit([w, nfbasis(w,1)]);
        modpr=nfmodprinit(nf, idealprimedec(nf,5)[1]);
        L = nfgaloisconj(w); vector(#L,i, nfeltreducemodpr(nf,L[i],modpr));
      (in this example, the last reduction is down from ~ 1minute to ~ 2s)
   32- make sure nfmodprininit chooses a monic T to represent
      Fq ~ Fp[X]/(T)  [#646]
   33- remove obsolete undocumented functions outerr, outbeauterr, outsor,
    outtex. Rename voir -> dbgGEN. Functions brute, outbrute, matbrute,
    outmat, sor, outbeaut are obsoleted and no longer documented.
   34- rename errfile -> pari_errfile, infile -> pari_infile, logfile ->
    pari_logfile
   35- extra test infrastructure [ drop file in src/test/in and possibly
    src/test/[32|64], then run Configure, no need to edit config/get_tests ]
   36- inline gerepileupto/gerepileuptoleaf + improve gerepileupto
   37- [libpari] cleanup of user / temp variable handling. manage_var
   obsoleted (kept for backward compatibility, to be removed), see Section 5.9
   for new equivalents. Fixes #633, #650
   38- fix t_LIST as components: v = [List()]; listput(v[1],) didn't work [#468]
   39- listcreate() and listkill() are obsolete, don't use them. L = List()
    should be enough in all cases. All lists now grow as needed, without
    requiring an awkward maximal length (from listcreate).
   40- rename sqred -> qfgaussred, signat -> qfsign,
    sqred1 -> qfgaussred_positive
   41- rename gaddmat -> RgM_Rg_add and swap arguments. Add RgM_Rg_add_shallow
   42- document library functions zv_neg, zm_transpose, fix typo in
    documentation for RgX_neg
   43- document library functions gmaxgs, gmaxsg, gmings, gminsg.
   44- document library function gfloor2n
   45- document library function zx_shift
   46- cleaned up splines handling [ ploth(,,,256) ] : remove quark_gen & QUARK
BA 47- implicitly local variables are lexically-scoped
BA 48- local and my can now appear anywhere in a program.
   49- [library] rename apell -> ellap
   50- Removed the OK_ULONG macro and renamed u_OK_ULONG -> SMALL_ULONG
   51- Rename BIGINT -> NO_VARIABLE [used by gvar()], VERYBIGINT -> LONG_MAX,
       MAXULONG -> LONG_UMAX. Remove BYTES_IN_LONG, TWOPOTBYTES_IN_LONG
   52- implement directly gsubsg [ was a macro calling a trivial wrapper ]
   53- optimize multiplication for quadratic t_POLMOD [ t_QUAD remains faster ]
   54- theta(q,z) very inefficient for large accuracies
   55- remove support for nf of the form [nf, change of variable] in
       nfnewprec() [#672]
   56- global() now obsolete and scheduled for removal
LGr57- 'make ctags' gctags-specific flags [ add #defines, typedefs ]
   58- prototype of nfsubfields [ 2nd argument GEN -> long ]
   59- allow t_FFELT in issquare / issquarerem / ispower
   60- sqrt(t_SER) now uses Newton iteration
   61- rename gissquarerem -> gissquareall, uissquarerem -> uissquareall,
       Z_issquarerem -> Z_issquareall (analogy with sqrtrem was faulty: we do
       not store a remainder but the square root)
BA 62- User functions are now regular variables holding values of type t_CLOSURE
   63- Output of \u is now sorted
   64- More explicit error messages in gp_history, e.g.
       ***   History result %10 not available [%1-%6]
BA 65- vecsmall_uniq(v) no longer assumes that v is sorted
   66- allow ellorder to handle curves over Fp, add an optional parameter to
       indicate a multiple of the order as in znorder.
   67- allow polcyclo(n, a) for an arbitrary a [ had to be a variable ] +
       major speedups: polcyclo(10^6) 5min -> 16ms. Similarly for polchebyshev,
       pollegendre and polhermite
   68- ?0 listed all "user-defined identifiers". Restrict to "functions"
   69- use simplify in setsearch() and Set() [#707]
   70- change gcd(t_POL, t_POL) so that inexact inputs have scalar gcd
       [ used to compute a "sensible" approximate gcd ]

  Added
    1- library functions expu, adduu, subuu
    2- library functions divisorsu, factor_pn_1 (using cyclotomic factors and
       Aurifeuille), merge_factor
    3- library function divru, dbllog2r
    4- library function ZX_gcd, ZX_isirreducible
    5- library function gtos
    6- library function pari_daemon
    7- library function Fp_sqr, padicsqrtlift
    8- library function RgXQ_inv
    9- bench 'extract'
   10- charpoly: Berkowitz division-free algorithm (new default) [#541]
   11- library function phiu
   12- library function mkvecsmalln
   13- library function chk_gerepileupto, dbg_gerepile, dbg_gerepileupto
   14- library function gen_sort_uniq, gen_indexsort_uniq+add flag 8 in vecsort
   15- library function remi2n, cmp_RgX, gen_cmp_RgX
   16- library function RgV_dotproduct, ZV_dotproduct, RgV_dotsquare, ZV_dotsquare, FpV_dotproduct, FpV_dotsquare
   17- library function FpX_Fp_add_shallow
   18- library function Q_pval
BA 19- new keyword "my" to declare lexically-scopped local variables.
   20- new GP function listpop()
   21- library function RgX_inflate, RgX_deflate, Flx_inflate, Flx_deflate, poldeflate
   22- library function Fp_neg, Fp_add, Fp_sub
   23- library function ugcd
   24- library function vecinv, vecmul, vecpow, vecdiv, vecmodii
   25- library function RgM_isidentity, RgV_isin
   26- library function nfnewprec_shallow, bnfnewprec_shallow, bnrnewprec_shallow
   27- library function Zp_issquare
   28- library function dvmdsBIL, divsBIL, remsBIL
   29- library functions gassoc_proto, map_proto_G, map_proto_GG, map_proto_GL,
       map_proto_lG, map_proto_lGG, map_proto_lGL [ were private with
       other names ]
   30- default factor_proven
   31- experimental -balloon option to gphelp
LGr32- defaults graphcolormap, graphcolors [runtime-defined colormap for plots]
LGr33- bit 4096 [ complex ] to ploth
   34- member function .p / .mod for t_PADIC
   35- same mnemonic flags to plotrecth as in ploth
BA 36- member function .pol, .mod and .p for t_FFELT
BA 37- New type t_CLOSURE to store GP functions
   38- Ability to use arbitrary comparison functions in vecsort()
   39- library functions closure_callgen1, closure_callgen2, closure_callgen,
       closure_callgenvec.
   40- GP function apply(), select()
   41- library functions FF_issquare, FpXQ_issquare, Flxq_issquare
BA 42- library function vecsmall_uniq_sorted
   43- bench 'ff', 'exact0'
   44- has_getenv.c
   45- library function FpX_oneroot
   46- library function gcmpX
   47- library functions ZV_sort, ZV_sort_uniq, ZV_search

  Removed
    1- obsolete function kbessel2 (now static). Removed optional flag in besselk
    2- Removed obsolete optional flag in ellap(), Mod() and gcd()
    3- Rename gen_search_aux -> gen_search and document it [ old gen_search
       disappears: use tablesearch() ]
    4- undocumented library functions gpolylogz, polylog, polylogd, polylogp,
       polylogdold. Use polylog0 with appropriate flags.
    5- undocumented library function ghell2. Use ellheight0 with appropriate
       flag.
    6- obsolete function log0; use glog
    7- obsolete undocumented library function mpdivz, polzagreel, RgX_powers
       (use RgXQ_powers)
    8- obsolete undocumented library function sqred3
    9- the emacs/ directory: PariEmacs is now distributed separately.
   10- obsolete functions ellap0, apell2
   11- obsolete header file pariport.h
   12- obsolete error codes paramer1, valencer1, accurer, caracer1
   13- removed obsolete system-specific #ifdefs (macintosh, alliant)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Done for version 2.4.1 (released 28/03/2007):
  Fixed
    1- qflll / qflllgram (t_MAT with t_FRAC entries) would not reduce to the
       integer case (--> insufficient precision, SEGV) [#505]
    2- [Cygwin] missing -L... -lgmp when compiling with gmp.
    3- ispower(522^3) -> 0 [ looked like a 7th power to is_357_power(), which
       then forgot to test for cubes ] [#506]
LGr 4- [gphelp] race condition --> incomplete cleanup (improved patch BA)
    5- Cleanup library linking: do not link libpari with -ld [only gp], do
    not link gp with -lgmp [only libpari]. Side effect: libgmp.so no longer
    needed for modules compiled by gp2c-run
    6- when nf.disc < 0, nf.diff was an incorrect PARI ideal [#510]
    7- nf.codiff was only correct up to multiplication by some rational number
       (a divisor of nf.disc) [#510]
    8- inaccuracy (>= 2ulp) in [cached] log(2) [#498]
    9- exp, sinh, asinh, tanh, atanh were inaccurate near 0
   10- [GMP kernel] forvec(x=[[-1,0]],print(x)) --> error [#509]
       [ 'resetloop' failed when passing through '0' ]
   11- nfbasistoalg(nfinit(y),x) created an invalid t_POLMOD
   12- incorrect result in ZX_resultant (accuracy loss in bound computation)
   13- bnfinit(): avoid further precision problems for large degree fields
   14- [Configure] gcc-specific flags were used on linux/freebsd/cygwin, even
       when __gnuc__ was unset
   15- factor( pure power FqX ) --> SEGV
   16- [GMP kernel] polrootsmod(f, 4) --> wrong result [ low level t_INT
       manipulation not using the int_* macros ]
   17- polrootspadic(f, 2, r) --> some roots would be found twice [ due to
       FpX_roots(f, 4) called ]  [#521]
   18- ??sumalt doesn't compile: in GPHELP, treat \ref in verbatim mode [#518]
   19- matinverseimage returned [;] when no pre-image exists. Conform to
       the docs: "an empty vector or matrix", depending on the input types.
   20- [Configure] abort when $CFLAGS is not supported by $CC
   21- 3.5 % 2 --> error [ should be 0.5 ]
   22- sin(1/10^100) --> 0e-28 [ also affected cos,tan,cotan ]
   23- fix e.eta and elleta such that e.eta = 2 ellzeta(e,e.omega/2)
       [ was ellzeta(e,e.omega/2) ]. COMPAT.
   24- elleta(e) was different from elleta(e.omega). Analogous problems
       in all quasi-elliptic functions. COMPAT: change e.omega so that
       e.omega[1] / e.omega[2] belongs to the Poincare half plane [ used
       to be the inverse ]. Together with 24: the Legendre relation now reads
       w1 e2 - w2 e1 = 2 I Pi
       Rationale: 1) the action of Sl_2(R) becomes the standard one, not a
       twisted one 2) fixes quite a few normalization problems in our code.
   25- check that k >= 0 in thetanullk [#531]
   26- isprime(-2,1) returned 1
   27- Fix 'Not enough precision in thue' error
BA 28- [OS X] Fix kernel detection on x86_64-darwin
   29- Remove "VERY long time" Warning in bnfcertify (few minutes nowadays)
BA 30- missing prototype for documented function ZY_ZXY_rnfequation
   31- sqrt(x^2/y^2) --> SEGV [#536]
   32- \r foo no longer worked if foo was a directory and foo.gp a valid
       input file [#540]
BA 33- [Configure] spectacular failure to recognize gcc under some locales.
   34- polredabs(x^8+2*x^6-5*x^4+78*x^2+9) was incorrect [ missed
       x^8+6*x^6-x^4+54*x^2+25 due to incorrect "skipfirst" ]   [#542]
   35- typo in resmod2n (both kernels) [#546]
   36- At \p28, 0.1 - 0.1 would return 0.E-30 instead of 0.E-29
BA 37- missing prototype for documented function FpX_div_by_X_x
   38- isprime(,0) very slow when primelimit is large [#546]
   39- nfmodprinit could create FpX's which were not reduced mod p
   40- O(x^3)^(1/2) was O(x^2) instead of O(x)
RB 41- the following TODO item:
       v = vector(2); v[1] = v = 0  --> SEGV. Occurs with high probability if
       any variable is "deleted", while it (or part of it) is still in use
       Reference count could be helpful here.
   42- substpol(x^-2+O(x^-1),x^2,x) --> error [#555]
BA 43- [TLS] addss, addsr and subsr were not reentrant.

  Changed
    1- concat(t_VECSMALL, t_VECSMALL) to return the concatenated vector
       [was: a vector with two t_VECSMALL entries]
    2- pariprintf() so that it handles t_STR as print() [ don't include quotes ]
LGr 3- [Makefile] make generated src/funclist independent of locale
    4- Extend Pocklington-Lehmer to the case N-1 = FU, F > N^(1/3)
BA  5- Much faster base-2 to base-10 conversion.
BA  6- FpX_Fp_add() is now clean.
BA  7- rename ZY_ZXY_resultant -> ZX_ZXY_resultant, ZY_ZXY_rnfequation ->
       ZX_ZXY_rnfequation and FpY_FpXY_resultant -> FpX_FpXY_resultant.
BA  8- FpV_polint() now take a variable number as last parameter.
    9- use Miller-Rabin-like improvement in znprimroot and FpXQ_gener
   10- indexrank, indexsort and indexlexsort now return t_VECSMALLs
   11- API for gen_sort, vecsort

  Added
BA  1- derivnum(x=a,expr) for numerical derivations
BA  2- library function strntoGENstr
    3- function Vecrev
BA  4- ppc64 level0 inline assembly kernel
    5- library function floor_safe()
    6- library function itostr()
BA  7- library function Fp_div(), Fp_mul()
BA  8- library function FpXQ_norm()
BA  9- library functions FlxX_resultant() and Flx_FlxY_resultant()
BA 10- library function FlxY_Flx_div()
BA 11- library function Flm_transpose()
BA 12- library function Flx_Fl_add()
BA 13- library function Flxq_div()
BA 14- function stirling (Stirling numbers of 1st and 2nd kind)
BA 15- library function FpX_valrem()
   16- library function Flxq_gener
BA 17- library function Flxq_norm, Flxq_minpoly, Flxq_charpoly
BA 18- [toplevel benchmark] ffisom
BA 19- library functions Fp_order, FpXQ_order, FpXQ_log, FpXQ_sqrtn,
       Flxq_order, Flxq_log, Flxq_sqrtn
   20- library functions gen_sort_inplace, gen_indexsort, sort_factor,
       indexvecsort
BA 21- New PARI type t_FFELT and support functions (FF_*) for finite field
       elements.
BA 22- functions ffgen, fforder, fflog, ffprimroot for finite field elements.

  Removed
    1- obsolete functions readexpr(), readexpr_nobreak()
    2- pariemacs support from Configure
    3- obsolete functions sindexsort, sindexlexsort, sindexrank

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Done for version 2.4.0 (released 06/10/2006):
  Fixed
    1- typo in remiimul: wrong result in a "failsafe" branch (hardly ever
       taken). May affect modular powering /Fp if p >> 10^1300 [#457]
       Also affects the gmp kernel, for p >> 10^5800.
    2- rare accuracy problem in bnfinit
       P=x^8-787856*x^6+232721637848*x^4-30546112988506688*x^2+1503204734505922286224;
       setrand(1974190693); bnfinit(P) --> non invertible matrix in gauss
    3- inconsistent return type in nffactormod [#460]
    4- the following TODO item
        bnfinit may give wrong answers because we cheat on the value of
        Bach's bound, using B := 0.3 log^2 D by default, where 0.3 should
        really be 12 (under GRH). If the prime ideals of norm <= B do not
        generate the classgroup, we may not detect it, and compute junk. Ex:
        * setrand(3); bnfinit(y^4 + 1190*y^2 + 1416100).cyc
          --> [8,2,2,2]. The correct structure is [8,4,2].
        * setrand(1414185642); bnfinit(y^4 + 635*y^2 + 403225).reg
          [twice the correct value]
        * setrand(867341586); bnfinit(y^4-y^3+6122*y^2+6121*y+37466641).gen[2]
          [is principal]. Group structure and regulator are correct!
       Retune check_bach()
IZ  5- remove bogus test for "external help" (= ??)
IZ  6- [OS/2]: stack check [prevent oo recursion] present but not enabled
IZ  7- [OS/2]: bogus [BUG] in 'program' bench (install pb not detected)
    8- gammah(1+O(3^5)) --> incorrect type in ggamd.
    9- invalid read in copy_leaf [ SEGV in some libc ]
BA 10- ellheight short help was accidentally truncated.
   11- substpol(1+O(x^(2*n)),x^2,x) --> 1+O(x) instead of 1+O(x^n) [#470]
   12- [OS X] only use -no-cpp-precomp with Apple cc
   13- divrem(x,x,y) --> [1/y*x, 0]
   14- (1+x)/(1-x)/(1+x)^2 not simplified [#472]
   15- typo in qflll: in rare cases (exact input+floating point computation+
       precision increase in last-but-1 step), the returned base change is not
       properly updated in last iteration --> basis not LLL-reduced
   16- simplification missed in div_scal_rfrac() [ #473 ]
   17- ispower(x^k, k) would answer 0 for some x and k in {3,5,7} [#476]
   18- content(t_MAT with exactly 1 col) gave a wrong result
   19- random(N) was not uniformly distributed in [0,N-1] (use a reject
       strategy instead of moding out) [#210]
   20- rare bug in red_montgomery (returning 0 with + sign, an incorrect object)
       [ polrootspadic(x^11+x,11,10) --> corrupts gen_0 ]
   21- qfbsolve(Qfb(1,2,10),5) --> [0,0; 0,0] instead of 0 [#479]
   22- ispower(0, n, &z) would not set z
   23- wrong result in conversion t_QUAD -> t_PADIC whenever disc  = 1 (4)
JJ 24- gaddgsz macro was wrong [#481]
   25- gener_Fl(p^k) can't handle k > 1, use gener_Zl instead [#480]
SC 26- [GMP] mp_set_memory_functions was called with an incompatible realloc
       function. [#484]
   27- raising a t_QFI or t_QFR to the power n would return a reduced form (OK)
       except when n = 1. Make it systematic.
   28- in rare cases (return 0, native kernel) int_normalize was not suitable
       for gerepileupto
   29- idealnorm(nf, t_POL) didn't work
AM 30- [TeXmacs] typo in texmacs_output: x --> <bluex> [#491]
   31- accuracy errors in bnfinit: setrand(3); bnfinit(x^2+999999999999971)
       [ use a failsafe version of gmod (modr_safe), and increase accuracy
         if necessary ]
   32- possible corruption of gen_2 in mynegi() [ hnflll ]
   33- factor(4/x) --> [2, 2; x -1]  (content not removed)
BA 34- bnrstark prototype code was non-standard.
   35- rnfkummer(,,degree) often found too many fields [#482]
   36- loss of accuracy in p-adic ellinit: wrong digits and spurious errors:
       E.g i = 5; ellinit([1, -1, 1, -1, -14]*(1+O(17^i)))
   37- missing GC in forvec(,,2)
BA 38- Interrupting GP could lead to a freeze. [#488]
   39- possible stack corruption in charpoly(,1)
   40- Ser(x) raised an error [#499]

  Changed
    1- moved cgiv, gtofp to inline kernel
    2- reduce amount of memory needed by APRCL
    3- remove CPP from Configure tests (not needed)
    4- allow arbitrary sequences (not only expressions) in sum, prod, etc.
       sum(i = 1, 2, 1;2) --> 4  [ was: error ]
    5- checked that no 2-strong pseudoprime passes BSPW primality test up to
       10^15 (use Galway's list, extending Pinch's 10^13)
    6- random() now uses Brent's XORGEN (replaces congruential linear generator)
       getrand() returns the FSR internal state array, used by setrand(). The
       latter no longer returns the input seed. (Initial patch by RR)
    7- for install(): parser code 'E' is now obsolete, use 'I'
    8- if issquare(x, &n) == 1, always set n to a square root of x
    9- if ispower(x, k, &n) == 1, always set n to a k-root of x
   10- issquare(t_QFI or t_QFR) now is an error. Use explicitly
       issquare(component(x,1)) if you relied(?) on the old behaviour
   11- [library] rename gener_Fp -> pgener_Fp, gener_Fl -> pgener_Fp
       [ contrary to 'gener', these assume that their argument is prime ]
   12- [library] rename cyclo -> polcyclo, subcyclo -> polsubcyclo,
       tchebi -> polchebyshev1, legendre -> pollegendre
   13- restrict the types allowed in gaffect and gaffsg to scalar types and
       vectors of such.
   14- vecextract(x, bitmask) much faster [ read bits, don't use shifts ]
   15- improve hyperu
   16- remove gp-specific signal handler. Use the one in libpari
   17- [library] pol_1 and pol_x are no longer global arrays but functions
   18- [GMP] was inefficient for small sizes (e.g. 1-word operands)
   19- optimized rectplothrawin & plot (remove gaffect, use gtodouble
       whenever possible). Coordinate computations should be faster and more
       accurate.

  Added
BA  1- function Flm_rank
BA  2- Add experimental --enable-tls Configure option and thread-local stack
       support. Warning: using this option break the ABI.
    3- centerlift(t_PADIC)
    4- pgener_Zp, pgener_Zl
    5- dvdiu, dvdis, dvdiu, dvdsi, affgr
    6- cgetipos, cgetineg, togglesign
RM  7- polhermite, polchebyshev2
    8- allow rnf.zk and rnf.nf
    9- [toplevel benchmarks] bnfisintnorm, quadclassunit, rnfkummer
   10- ZX_factor, QX_factor
   11- allow gzip'ped elldata & galdata files
   12- allow t_VECSMALL in vecmin / vecmax
BA 13- add pari_stackcheck_init() function to control deep recursion detection
   14- Configure --tune flag

  Removed
    1- Odos/Makefile (no longer functional, obsolete)
    2- old CodeWarrior-specific hack (malloc)
    3- error code: intger2, affer3, overwriter
    4- useless static function gp_handle_SIGINT
    5- functions changevar() [ use substvec ] and reorder() [ use variable()
       to get the list of user variables ], global arrays pol_1[], pol_x[]
       ordvar[] and polvar[]