Changes commited to PARI GIT during the last two weeks

hyperellcharpoly(Mod(1,13)*(x+1)) -> fix SEGV [#2720]

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

remove space at end of line

1	1	doc/usersch5.tex

fix primecertisvalid false positive [#2723]

1	0	CHANGES
1	0	src/basemath/ecpp.c
2	1	src/test/32/isprime
1	0	src/test/in/isprime

document semantic of gequal0(t_COMPLEX)

5	1	doc/usersch5.tex

Make a few FpE_* / FpEQE_* functions suitable for gc_upto

FpJ_neg, FpE_to_FpJ, FpE_net, FpXQE_neg: replace return mkvecX(...) by
retmkvecX(...)

5	5	src/basemath/FpE.c

a_posteriori_errors: reduce variable scope

2	2	src/basemath/rootpol.c

gabs calls: fix incorrect (legacy) 'prec' argument

1	1	src/basemath/grossenchar.c
2	2	src/basemath/rootpol.c

ZC_elemz: do last step inplace too

47	39	src/basemath/hnf_snf.c

spurious return

1	1	src/basemath/hnf_snf.c

4- charpoly(matrix over p-adic fields): use Berkowitz in all cases

Much stabler than Hessenberg form, not much slower.

TODO: A _fast variant could replace computation with t_PADICs or
t_POLMOD over Qp with fixed-point arithmetic using the modular kernel

2	0	CHANGES
0	1	src/basemath/alglin2.c
7	1	src/test/32/charpoly
14	0	src/test/in/charpoly

qfminimize: division by zero, thanks Samuel Li [#2721]

1	0	AUTHORS
1	0	CHANGES
1	1	src/basemath/qfsolve.c
2	1	src/test/32/qfsolve
1	0	src/test/in/qfsolve

fix comment

1	1	src/basemath/ellpadic.c

14- bnfinit: typo making Grenie-Molteni's residue bound larger [#2706]

1	0	CHANGES
4	3	src/basemath/buch2.c

13- ellpointtoz(E/Qp, P) accuracy issues

- more accurate that requested: faulty use of gprec() instead of gcvtop
- failures with division by 0 in low accuracy (catastrophic
  cancellation): restart at higher accuracy

1	0	CHANGES
21	10	src/basemath/elltrans.c
14	15	src/test/32/ell
2	0	src/test/in/ell

ZX_to_padic: return p-adic 0 for 0 pol

3	1	src/basemath/hyperell.c

add test

2	1	src/test/32/ell
2	4	src/test/in/ell

bench typo

2	2	src/test/32/ell
1	1	src/test/in/ell

Fix Qp_descending_Landen: f5d7bb5124 was too optimistic

confusion between v(r) and v(R) = v(r) - vA: R = r/p^vA decreases quadratically
but we are handling r and must adjust by vA.

12	8	src/basemath/ellpadic.c
7	7	src/test/32/ell

rewrite get_R in terms of (exported) Buch_check

10	17	src/basemath/buch1.c
5	4	src/basemath/buch2.c
1	0	src/headers/paripriv.h

hyperellcharpoly: do not use carberkowitz with t_PADIC [#2720]

1	0	CHANGES
76	9	src/basemath/hyperell.c
2	0	src/test/32/hyperell
3	0	src/test/in/hyperell

hyperellisisom_gen: handle degree drop correctly

2	2	src/basemath/hyperell.c

mathnf: use in-place operations

116	70	src/basemath/hnf_snf.c

get_R: gtodouble could overflow for really large discriminants

3	4	src/basemath/buch1.c

get_row: do not use approximation of rational roots. [#2719]

E=ellinit([1,0,0,-8519572288638382835217723088514971910366397572001198415734563942679320,-302673780988173543152422516593707329574281613842246182404050555431591014586242532196838077153123840536000]);
localbitprec(64);ellrank(E)
 *** ellrank: bug in ellrank, wrong nfeltsign, please report.

1	0	CHANGES
1	1	src/basemath/ellrank.c

Add Configure test for sigsetjmp for 20th century OS like mingw

1	0	config/get_libc
11	0	config/has_sigsetjmp.c
4	0	config/paricfg.h.SH
9	0	src/headers/parisys.h

setrand(2); hyperellisisom(x^5+x+1, x^5+x+1) ->SEGV

1	0	CHANGES
1	0	src/basemath/hyperell.c
1	1	src/test/in/hyperellauto

BA   3- [macos] use sigsetjmp instead of setjmp for portability [#2715]

1	0	CHANGES