Changes commited to PARI GIT during the last two weeks

36- [libpari] QM_image, QM_image_shallow

1	0	CHANGES
6	0	doc/usersch5.tex
8	3	src/basemath/alglin1.c
3	10	src/basemath/modsym.c
2	0	src/headers/paridecl.h

Flxq_autpow: fix support for Flx_get_red

2	2	src/basemath/Flx.c

improve documentation

22	0	src/functions/number_theoretical/divisors
9	13	src/functions/programming/fordiv

34- M=mfinit([11,2,Mod(3,11)]); f=mfbasis(M)[1]; lfuncheckfeq(lfunmf(M,f))

-> type error. Embed everything in sight.

2	0	CHANGES
5	4	src/basemath/mftrace.c
2	1	src/test/32/mf
2	0	src/test/in/mf

algisassociative: accept input with first matrix not identity

14	4	src/modules/algebras.c
3	7	src/test/32/algebras
2	3	src/test/in/algebras

33- Flx_div_by_X_x(A, x) didn't handle deg(A) <= 0

1	0	CHANGES
7	3	src/basemath/Flx.c

update 32bit bench

1	1	src/test/in/nfields

23- removed obsolete function 'bnfcompress'

1	0	CHANGES
9	0	COMPAT
0	3	doc/refcard-nf.tex
1	1	src/basemath/base1.c
0	116	src/basemath/buch2.c
0	32	src/functions/number_fields/bnfcompress
1	1	src/gp/whatnow.h
0	1	src/headers/paridecl.h
2	5	src/test/32/compat
1	13	src/test/32/nfields
1	4	src/test/32/rnf
1	5	src/test/in/nfields
0	1	src/test/in/rnf

whitespace edit

1	1	src/basemath/buch2.c

update bench

1	1	src/test/in/nfields

lfunzeros(1,[-20,20]): zeros repeated twice

1	1	src/basemath/lfun.c
3	1	src/test/32/lfun
1	0	src/test/in/lfun

cleanup class_group_gen

12	12	src/basemath/buch2.c

simplify class_group_gen: no longer try to replace I by red(I^(-1))

11	55	src/basemath/buch2.c
13	13	src/test/32/bnrisgalois
2	2	src/test/32/nfields
7	5	src/test/32/rnfkummer
1	1	src/test/in/nfields

misc. cleanups

7	12	src/basemath/buch2.c

nfisincl: fix variable leak

1	1	src/basemath/base1.c

Flxq_autpow, FpXQ_autpow: use fold + precomputation

6	3	src/basemath/Flx.c
6	3	src/basemath/FpX.c

New test export

11	0	src/test/32/export
16	0	src/test/in/export

realroots: uniformize use of evenh [remove it]

11	12	src/basemath/rootpol.c

kill (spurious) warning

3	3	src/basemath/rootpol.c

RgX_gtofp_bit: remove useless (and ineffective) security

1	5	src/basemath/rootpol.c

cleanup lfunzeros

19	21	src/basemath/lfun.c

update

4	11	src/functions/l_functions/lfunzeros

22- lfunzeros: allow intervals not contained in R^+

1	0	CHANGES
64	49	src/basemath/lfun.c
12	12	src/functions/l_functions/lfunzeros
8	1	src/test/32/lfun
4	0	src/test/in/lfun

34- [libpari] ZX_sturm_irred, ZX_realroots_irred

1	0	CHANGES
8	0	doc/usersch5.tex
3	3	src/basemath/base1.c
3	2	src/basemath/mftrace.c
114	52	src/basemath/rootpol.c
3	1	src/headers/paridecl.h
3	3	src/modules/thue.c

export + local -> SEGV

f(i)=local(z);i^2+1
export(f);
parfor(i=1,10,f(i))

1	0	CHANGES
4	2	src/language/eval.c

improve doc

7	3	src/functions/l_functions/HEADER
23	21	src/functions/l_functions/lfuncreate

lfuncreate: add support for closures of arity 0

1	0	CHANGES
10	0	src/basemath/lfunutils.c
27	0	src/functions/l_functions/lfuncreate
1	1	src/headers/paripriv.h
23	1	src/test/32/lfuntype
2	1	src/test/in/lfuntype

lfuninit: call ldata_newprec to increase the precision when needed

8	2	doc/usersch8.tex
6	4	src/basemath/lfun.c
20	0	src/basemath/lfunutils.c
2	1	src/headers/paridecl.h
2	2	src/test/32/lfuntype

New libpari function closure_callgen0prec

1	1	CHANGES
3	0	doc/usersch5.tex
1	0	src/headers/paridecl.h
12	0	src/language/eval.c

avoid overrunning diffptr in compute_invresquad [#2175]

1	0	CHANGES
7	6	src/basemath/buch1.c

tune: make dichotomic search the default

17	8	src/test/tune.c

ZX_Uspensky: rename variables

26	37	src/basemath/rootpol.c

change ZX_Uspensky so that it finds all rational roots first

- negligible cost
- this is a often a big speed improvement
- guarantees that all rational roots are found

1	0	CHANGES
9	7	doc/usersch5.tex
54	104	src/basemath/rootpol.c

tune -d: fix potential infinite loop

1	1	src/test/tune.c

cleanup usp: remove deg0

7	15	src/basemath/rootpol.c

cleanup ZX_Uspensky

40	46	src/basemath/rootpol.c

cleanups in ZX_Uspensky and realroots

12	25	src/basemath/rootpol.c

FlxqM_mul_Kronecker: use get_Flx_degree instead of degpol [#2174]

1	0	CHANGES
1	1	src/basemath/Flx.c
1	0	src/test/32/mat
4	0	src/test/in/mat

FlxqX_safegcd: use Flx_get_red

1	0	src/basemath/Flx.c

Flx_addshift: handle x==0 correctly for Karatsuba

1	1	src/basemath/F2x.c
2	2	src/basemath/Flx.c

tune: remove pseudo types t_Fqx, t_Fhx

5	19	src/test/tune.c

ZM_inv_i: make sure denominator is positive [#2173]

1	0	src/basemath/alglin1.c
4	0	src/test/in/nf

compatlift: lift(t_POLMOD) can return a t_INT

4	1	src/basemath/mftrace.c

New libpari function RgV_type2

1	0	CHANGES
4	0	doc/usersch5.tex
1	1	src/basemath/bibli2.c
12	0	src/basemath/polarit2.c
1	0	src/headers/paridecl.h

20- asymptotically fast variant for RgX_translate algorithms

1	0	CHANGES
18	2	src/basemath/RgX.c

direuler_bad: do not use degpol for a t_SER

3	2	src/basemath/dirichlet.c

FlxqX_*: use FlxX_shift and not RgX_shift (which uses gen_0)

9	7	src/basemath/Flx.c

Flx.c: move Flx_equal to Flx section

10	10	src/basemath/Flx.c

conformal_pol: avoid conjugation if a is real

3	1	src/basemath/rootpol.c

Flx: Update tuning parameters

22	22	src/desc/deftune
36	36	src/headers/paritune.h
43	43	src/kernel/gmp/tune.h
18	18	src/kernel/none/tune-gen.h
42	42	src/kernel/none/tune.h

ellweilpairing(E,P,P, odd m) -> wrong result [#2172]

1	0	CHANGES
2	1	src/basemath/F2xqE.c
2	1	src/basemath/FlxqE.c
4	1	src/test/32/ellweilpairing
7	0	src/test/in/ellweilpairing

mygprecrc: avoir useless copies

8	5	src/basemath/rootpol.c

Flx.c: fix usage of tuning parameters

9	9	src/basemath/Flx.c

vecperm_extendschreier: off-by-1 error

1	1	src/basemath/perm.c

asymptotically fast conformal_pol + rewrite scalepol in terms of RgX_rescale

30	19	src/basemath/rootpol.c

maxlengthcoeffpol: removed, replaced by maxbitcoeffpol

0	15	src/basemath/Flx.c

improve diagnostics

3	3	src/basemath/RgX.c
1	1	src/basemath/rootpol.c

all_roots: improvements

- don't take leading_coeff into account in initial precision guess
- simplify handling of P(x^h)

10	13	src/basemath/rootpol.c

tune: make sure DFLT_mod/DFLT_mod2 are prime

2	2	src/test/tune.c

deftune: resort using C locale

18	18	src/desc/deftune

galois_analysis.group: remove broken/obsolete type cast

1	1	src/basemath/galconj.c

Flx_mul/Flx_sqr: rationalize tuning parameters

We keep:
XXX for SMALL_ULONG, XXX2 for larger numbers

51	97	src/basemath/Flx.c
18	25	src/desc/deftune
28	42	src/headers/paritune.h
36	50	src/kernel/gmp/tune.h
14	21	src/kernel/none/tune-gen.h
36	50	src/kernel/none/tune.h
25	69	src/test/tune.c

struct galois_analysis: ga.group should be a long

1	1	src/basemath/galconj.c

galoisinit: do not use new algorithm for small groups

7	3	src/basemath/galconj.c
9	4	src/test/32/bnrclassfield
6	6	src/test/32/bnrisgalois
2	2	src/test/32/idealramgroups
8	7	src/test/32/stark
1	2	src/test/in/bnrclassfield

gammamellininv: add sanity check for Vga = []

3	2	src/basemath/mellininv.c
7	1	src/test/32/gammamellininv
4	0	src/test/in/gammamellininv

final cleanup Klargeinit

11	9	src/basemath/mellininv.c

simplify vp

7	4	src/basemath/mellininv.c

cleanup Klargeinit

56	43	src/basemath/mellininv.c

32- [libpari] Vgaeasytheta

+ factor out get_SM from Klargeinit0

1	0	CHANGES
3	0	doc/usersch8.tex
8	8	src/basemath/lfun.c
30	20	src/basemath/mellininv.c
1	0	src/headers/paridecl.h

improve Klargeinit0

6	6	src/basemath/mellininv.c

improve Klargeinit0

35	30	src/basemath/mellininv.c

SEGV in mpqs at \g6 and above [#2170] [#2171]

2	2	src/modules/mpqs.c

rootpol.c: remove karasquare/cooksquare by ZiX_mul based on ZX_mul

33	261	src/basemath/rootpol.c

lfunhardy: fix support for negative t

1	0	CHANGES
1	1	src/basemath/lfun.c
1	0	src/test/32/lfun
1	0	src/test/in/lfun

add tests

4	1	src/test/32/lfun
3	1	src/test/in/lfun

change SPLIT tuning. Fixes [#2168]

1	1	src/basemath/buch2.c

remove unused variable

1	1	src/basemath/buch3.c

fix lie / typo in primecertify [from #2168 but doesn't fix that bug]

2	3	src/basemath/buch3.c
2	0	src/test/in/nf

improve / fix ZM_pivots and ZM_gauss

- 19- improved ZM_rank for rectangular matrices (transpose + use fewer primes)
- 27- ZM_gauss returned incorrect result for t_MAT RHS and 0 input

2	0	CHANGES
9	10	src/basemath/alglin1.c

eisker: retune use_Petersson, treat + part correctly (faster)

half the rows can be removed before kernel

19	32	src/basemath/modsym.c
1	1	src/test/32/modsym

solve: test solvestep

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

solvestep was using too much stack

5	0	src/language/sumiter.c

kill warnings

3	2	src/basemath/Flx.c

unused variable

0	1	src/basemath/Flx.c

Flx_mul/Flx_sqr: allow interpolation on arbitrary bit length

78	14	src/basemath/Flx.c

Flx_translate1: retune for 32-bit archs

15	5	src/basemath/Flx.c

fix and retune Flx_translate1

47	22	src/basemath/Flx.c

ZM_inv_i: make sure returned matrix is primitive [#2169]

1	2	src/basemath/alglin1.c
3	0	src/test/in/nf