Changes commited to PARI GIT during the last two weeks

39- bnfinit: SEGV due to precision issues [#1592]

V=Vecsmall([-7965853153201475372,3569703424444014425,-793932311338574722,-763572 6163360422296,2439300824977180422,6979578556528110283,9084417258731406046,-35121 87990227595876,-3657147821972591116,2569005965964251197,3949088045253347112,-117 0834773626916202,-2968887346774536399,6295681545925013733,-2573709607006174214,- 8381721930084359707,4574676558936321425,-8703490018498810351,4022996495683334358 ,-7035763370579912875,-2755734279867978087,-3367292733493080150,-509948418885793 6393,1406562832059871170,-8867790548553449467,4217714981497738961,-3378421154341 451369,9034669664584953901,-3005482072524959369,-2628326000794931590,-3819214751 934038261,413237145142673039,-1760137937535919478,1855032851225718598,2649487062 279305629,-5696436474525086295,-5833942848832265282,7819383707568402483,10698688 95849977255,-343949684743301590,-5388415652812096641,1997476480889124605,-308878 3767981915338,3116335093600988454,8956184277465432916,-2657541847204568646,-6609 147374740248013,4470684606703096915,-2647448632025672283,3258063797902873231,-90 4655602857655266,2785571539627046683,4256664021139120350,7199241604668107400,-29 20824926362980543,2328249437499564229,-3673843003078560168,807011202107076158,37 60483255587252662,-6218291806396196933,5296538376412447968,-8557559671731162751, 813135531532701414,-2217330915153714271,11,-7322758914588968699]);
setrand(V);bnfinit(x^3-42751*x^2-20250*x-29251)

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

OCgerepile: handle correctly !isonstack(x)

10	7	src/language/eval.c

Update help test

2	2	src/test/32/help

Add GP function powers

1	1	CHANGES
2	0	src/basemath/trans1.c
6	0	src/functions/operators/powers

fix roundr & variants

install(roundr_safe,G)
roundr_safe(-4.499999999997613659) -> -5 in 32bit at default accuracy

44	29	src/basemath/gen3.c
13	13	src/test/32/compat
2	2	src/test/32/idealramgroups
13	13	src/test/32/linear
2	2	src/test/32/nf
25	25	src/test/32/rnfkummer
13	13	src/test/64/compat
2	2	src/test/64/nf
9	9	src/test/64/rnfkummer

CHANGES: mention fix for divrr rounding

1	1	CHANGES

gen_powers: fix one

2	2	src/basemath/trans1.c

Add PARI function gpowers

1	0	CHANGES
6	1	doc/usersch5.tex
8	0	src/basemath/trans1.c
1	0	src/headers/paridecl.h

Fq_sqrtn: use proper fix, previous was not enough

6	2	src/basemath/polarit3.c

38- Fq_sqrtn(t_INT,..,&zeta) could return a wrong root of 1

+ fix doc of all modular sqrtn functions

1	0	CHANGES
15	20	doc/usersch5.tex
2	3	src/basemath/polarit3.c

resultant_all: fix doc + comment

2	2	doc/usersch5.tex
1	1	src/basemath/polarit2.c

doc typos

5	4	doc/usersch5.tex

24- renamed ZV_gcdext -> ZV_extgcd for consistency with other gcdext methods

1	0	CHANGES
1	0	COMPAT
2	2	doc/usersch5.tex
2	2	src/basemath/hnf_snf.c
1	1	src/headers/paridecl.h

ellL1_bitprec: reinstate shortcut L(1) = 0 when sign = -1

1	0	src/basemath/ellanal.c

\d no longer printed default

fix help message class 16 -> 17 (due to the intoduction of section ?6)

2	2	src/language/default.c

test-incgam: make test independent of precision

7	3	src/test/32/incgam
0	73	src/test/64/incgam
1	1	src/test/in/incgam

test-nf: use default(realprecision)

1	2	src/test/32/nf
1	1	src/test/in/nf

fix accuracy loss in make_M

avoid losing accuracy by pre-inverting denominators, divite by a t_INT instead

3	6	src/basemath/base1.c
122	123	src/test/32/compat
51	52	src/test/32/member
3	3	src/test/32/nf
32	32	src/test/32/nfields
5	5	src/test/32/zetak
129	129	src/test/64/compat
1	1	src/test/64/incgam
44	50	src/test/64/member
4	4	src/test/64/nf
35	35	src/test/64/nfields
15	15	src/test/64/rnfkummer

incgam_0: use enough accuracy (add 10 guard bits)

1	1	src/basemath/trans3.c
2	2	src/test/32/incgam

Buchquad: use input accuracy for t_QFR component

instead of impredictable PRECREG

2	2	src/basemath/buch1.c
2	2	src/test/32/compat
3	3	src/test/32/member
2	2	src/test/32/nfields
2	2	src/test/64/compat
3	3	src/test/64/member
2	2	src/test/64/nfields

improve final reduction in rnfkummer

use Babai's algorithm instead of reducebetanaive + make sure to use enough
accuracy in floating point computations (use get_arch_real)

0	2	src/basemath/bibli1.c
1	0	src/headers/paridecl.h
42	74	src/modules/kummer.c
130	129	src/test/32/rnfkummer
121	125	src/test/64/rnfkummer

make test-kernel kernel-independent

1	1	config/Makefile.SH
19	19	src/test/32/kernel
0	23	src/test/64/kernel
5	4	src/test/kerntest.c

Remove src/test/64/{ellanal,bnrisgalois} (identical to 32bit)

0	9	src/test/64/bnrisgalois
0	24	src/test/64/ellanal

Fix test-kernel (add new_chunk_resize)

1	0	src/test/kerntest.c

divrr(x,y), lg(y) = 3: incorrect rounding

9	2	src/kernel/gmp/mp.c
9	2	src/kernel/none/mp.c
3	3	src/test/32/rnfkummer
26	25	src/test/64/compat
3	3	src/test/64/nfields
7	7	src/test/64/rnfkummer

cleanup filtre0

- inline check_filtre (uselessly exported)
- always initialize F->buf (to NULL if no supplied output buffer)
- check whether we are in "help context" (following a ?) in a more
  transparent way

The last two items fix: eval("=") => SEGV [#1591]

0	1	src/headers/paripriv.h
27	21	src/language/es.c
3	0	src/test/32/eval
2	0	src/test/in/eval