Changes commited to PARI GIT during the last two weeks

116- thue(thueinit(x^3-3*x^2+1,1), 2187) -> domain error in mplog [#2397]

1	0	CHANGES
1	1	src/modules/thue.c
2	1	src/test/32/thue
2	0	src/test/in/thue

classgp_make_pcp: further cleanups

- primeform(D, 1) is already reduced;
- primeform_u is not documented to work with p = 1 (it happens to work,
  in an inefficient way)

7	10	src/basemath/polclass.c

fix GC in qfr_1_by_disc

5	3	src/basemath/Qfb.c

rewrite / improve qfr_1_by_disc

13	10	src/basemath/Qfb.c

cleanup classgp_make_pcp

18	28	src/basemath/polclass.c

100- bnr_char_sanitize, bnr_subgroup_sanitize

1	0	CHANGES
14	0	doc/usersch6.tex

Revert "improve compatibility gcharlocal/gcharidentify"

Reverts eea62398a and allow t_VEC in gcharidentify.

5	2	src/basemath/grossenchar.c
2	0	src/functions/number_fields/gcharidentify
1	1	src/functions/number_fields/gcharlocal
3	3	src/test/32/gchar

make lfunvgasub static

5	2	src/basemath/hgm.c
1	1	src/basemath/lfunutils.c
0	1	src/headers/paridecl.h

add test

2	1	src/test/32/prime
1	0	src/test/in/prime

_uisprime: delete dead code

6	51	src/basemath/prime.c

idealsqrtn(A): improve the silly case A in Z + add test

13	8	src/basemath/base4.c
2	1	src/test/32/ideal
1	0	src/test/in/ideal

add test

2	1	src/test/32/ideal
1	0	src/test/in/ideal

add gchareval to refcard

1	1	doc/refcard-nf.tex

rewrite quadunit for non-maximal orders + make quadunit_basecase static

145	84	src/basemath/quad.c
0	1	src/headers/paridecl.h

typo

1	1	src/functions/programming/forsquarefree

115- colors = "no" (or "") handled incorrectly [#2402]

1	0	CHANGES

sd_colors: handle v = "no" (or "") properly

6	6	src/language/default.c

further numerical unstability in hypergeom [#2382]

14	6	src/basemath/hypergeom.c
6	1	src/test/32/hypergeom
17	3	src/test/in/hypergeom

114- numerator(1/x+1/y, 'x) -> x*y

1	0	CHANGES
1	1	src/basemath/gen3.c
2	0	src/test/32/content
3	0	src/test/in/content

99- new libpari functions ZV_ZV_mod (aka vecmodii), ZM_ZV_mod

reduce the matrix output by bnrsurjection

2	1	CHANGES
8	0	doc/usersch5.tex
5	12	src/basemath/base3.c
2	2	src/basemath/bnflog.c
6	8	src/basemath/buch3.c
1	1	src/basemath/char.c
1	5	src/basemath/hnf_snf.c
1	1	src/basemath/mftrace.c
51	49	src/headers/paridecl.h
10	0	src/headers/pariinl.h
1	1	src/modules/stark.c
2	2	src/test/32/bnrmap

whitespace edit

1	4	src/basemath/buch3.c

doc typo

2	2	doc/usersch5.tex

99- new libpari functions abmap_kernel, abmap_subgroup_image

1	0	CHANGES
33	15	doc/usersch6.tex
8	8	src/basemath/buch3.c
2	2	src/headers/paridecl.h
6	6	src/modules/stark.c

evaluation of gchar: chareval -> gchareval

1	4	src/basemath/char.c
3	3	src/functions/number_fields/gcharduallog
24	0	src/functions/number_fields/gchareval
4	4	src/functions/number_fields/gcharidentify
1	1	src/functions/number_fields/gcharlocal
1	1	src/functions/number_fields/gcharlog
0	7	src/functions/number_theoretical/chareval
15	8	src/test/32/gchar
2	2	src/test/32/gchar-large
38	27	src/test/in/gchar
6	6	src/test/in/gchar-large
30	31	src/test/in/gchar-lfun

section gchar in libpari doc

12	0	doc/usersch6.tex

test gcharidentify(gcharlocal)

2	1	src/test/32/gchar
3	0	src/test/in/gchar

improve compatibility gcharlocal/gcharidentify

It is convenient to be have (for suitable Lv)

chi == gcharidentify(gc,Lv,[gcharlocal(gc,chi,v)|v<-Lv])

gcharlocal used to output [theta] for a finite place not dividing the
modulus, whereas gcharidentify expects a single value at those places.

One can
- either remove the length 1 t_VEC in gcharlocal
- or ask the user to bracket each character value in gcharidentify.
Fixing the tests convinced me the first option is preferable.

2	3	src/basemath/grossenchar.c
1	1	src/functions/number_fields/gcharlocal
4	4	src/test/32/gchar

cleanup alg_pmaximal

in particular declaration not at beginning of block (not allowed in C89)

10	15	src/modules/algebras.c

43- faster implementation of binomial

1	0	CHANGES
60	5	src/basemath/bibli2.c
3	1	src/test/32/binomial
2	0	src/test/in/binomial

algebras: add GC in alg_pmaximal

4	2	src/modules/algebras.c

minor improvement in binomial

2	2	src/basemath/bibli2.c

BA  98- rnfabelianconjgen

1	0	CHANGES
12	2	doc/usersch6.tex

move rnfisabelian to nffactor.c + cleanup rnfabelianconjgen

0	12	src/basemath/buch3.c
34	22	src/basemath/nffactor.c
1	1	src/headers/paridecl.h

parse.y: silence warning about pari_nerrs not being used.

1	1	src/language/parse.y

get_maxord: test for pseudoprimality *after* cheaper test

2	2	src/basemath/base2.c

gen_pow_fold: better diagnostic in GC

1	1	src/basemath/bb_group.c

nfmaxord even lazier factorization

When a composite is found among the "primes", no need to factor it
completely: only need to find a true prime divisor and relaunchthe algorithm

17	3	src/basemath/base2.c
2	1	src/test/32/nf
2	0	src/test/in/nf

Remove unused variables

2	3	src/basemath/RgX.c
1	3	src/basemath/ellisog.c

get_red_G: remove dummy variable i

2	2	src/basemath/base1.c

parisys: Remove ANYARG, not used anymore

0	2	src/headers/parisys.h

eval.c: Improve EVAL_f macro to cast to the right prototype.

27	26	src/language/eval.c

forfactored & friends use proper "constant" when determining whether to sieve

2	1	CHANGES
3	0	doc/usersch5.tex
4	4	src/basemath/ifactor1.c
1	0	src/headers/paridecl.h
9	4	src/language/sumiter.c

improve forfactored variants over short intervals [#2398]

52	7	src/language/sumiter.c

znprimroot: error when (Z/n)^* non-cyclic

The function is only used from GP and we must factor n and p-1 anyway,
it's not going to slow down anything critical.

4	3	src/basemath/arith1.c
2	2	src/functions/number_theoretical/znprimroot
2	2	src/test/32/err
8	2	src/test/32/zn
2	0	src/test/in/zn

whitespace edit

1	2	src/basemath/polclass.c

refcard: add factormodcyclo

1	0	doc/refcard.tex
1	1	doc/refmacro.tex

src/basemath/polclass.c: force k=2 on 64bit systems

A good example is
polclass(-23512271, 1)
On 32bit systems, there is not enough primes available.

43	37	src/basemath/polclass.c

update nflist doc

11	13	src/functions/number_fields/nflist

fix GC problem in polclass_roots_modp: don't destroy vshape

2	3	src/basemath/polclass.c
2	1	src/test/32/polclass
2	0	src/test/in/polclass

polclass: classgp_make_pcp: initialize G to 0

Ideally we would only keep what we need.

8	8	src/basemath/polclass.c

Use new multiply instructions for MIPS R6

MIPS R6 Removed Hi,Lo registers and introduced new instructions
to multiply targeting GPR, use those instructions to workaround
build failure.

Signed-off-by: Jiaxun Yang 

22	0	src/kernel/mips/asm0.h
22	1	src/kernel/mips64/asm0.h

doc typos

1	1	src/functions/sums/solve

cleanup

4	4	src/basemath/pclgp.c