Changes commited to PARI GIT during the last two weeks

fix inefficiency in polint_chinese

often called with mA = [;]. From test-mat on 'pari':
  p = nextprime(2^63);
  q = nextprime(p + 1);
  {
     forvec(v = vector(8, i, [0, 2]),
            matker([p^v[1]*q^v[2], p^v[3]*q^v[4];
                    p^v[5]*q^v[6], p^v[7]*q^v[8]], 1));
  }
Before:
cpu time = 43,088 ms, real time = 15,258 ms.

After:
cpu time = 972 ms, real time = 863 ms.

9	5	src/basemath/arith1.c

17- u_forprime_next around 2^63 could return primes larger than upper bound

+ add a few more primes

1	0	CHANGES
24	13	src/language/forprime.c

update bench

1	1	src/test/in/linear

improve ZM_ker_i

- avoid computing contents twice
- return matrix with integer entries [ changes output of QM_ker ]

6	6	src/basemath/alglin1.c
4	4	src/test/32/linear
5	5	src/test/32/mat
1	1	src/test/in/linear

bench mat: move errors at end of file

35	35	src/test/32/mat
13	12	src/test/in/mat

lfuninit_vecc: reduce memory usage

- gammamellininvrt is no longer GC-clean

2	1	doc/usersch8.tex
8	4	src/basemath/lfun.c
2	4	src/basemath/mellininv.c

18- [libpari] radicalu

1	0	CHANGES
1	0	doc/usersch5.tex
0	7	src/basemath/galconj.c
7	0	src/basemath/ifactor1.c
2	3	src/basemath/mftrace.c
1	0	src/headers/paridecl.h

fix ploth typo

1	1	src/graph/plotport.c

fix rnfkummer bench (# results on 32 and 64bit)

1	10	src/test/32/rnfkummer
2	1	src/test/in/rnfkummer

improve doc

4	1	src/functions/polynomials/polroots

16- nf2 = nfnewprec(nf) could return incompatible nf2 [#2154]

polroots(nf.pol) could be sorted differently depending on the precision

1	0	CHANGES
6	3	src/basemath/buch2.c
7	9	src/basemath/rootpol.c
10	1	src/test/32/rnfkummer
0	2	src/test/in/rnfkummer

sprkinit: change GEN gk to long k

6	6	src/basemath/base3.c

idealprincipalunites_i remove useless diagnostic

(confusing when called from sprkinit)

0	2	src/basemath/base3.c

cleanup bnrgaloismatrix

8	9	src/basemath/buch3.c

simplify parploth documentation

29	95	src/functions/programming/parploth

improve F2x_issmooth_squarefree

13	7	src/basemath/F2x.c

use F2xq_mul + whitespace edits

4	7	src/basemath/F2x.c

15- conjvec(t_FFELT of characteristic 2) was missing a conjugate

1	0	CHANGES
1	1	src/basemath/F2x.c
8	8	src/test/32/ff

plotrecthin: no need for snm_closure

1	2	src/graph/plotport.c

uniformize gen_parapply / paralell code in plotrecthin

6	5	src/basemath/alglin3.c
7	5	src/graph/plotport.c

fix parploth, broken in 1e4769a0d

3	10	src/graph/plotport.c

cleanup plotrecthin

16	17	src/graph/plotport.c

whitespace edit

0	2	src/graph/plotport.c

plotrecthin: simplify parallel support

separate acquisition of N "coordinates" (eval) from interpretation
(filling dblPointList).

39	69	src/graph/plotport.c

shorten variable names

40	41	src/graph/plotport.c

fix undefine behaviour in polzag1 [ left shift of negative value ]

+ whitespace edit

3	3	src/language/sumiter.c

mpqs_eval_sieve: change variable name

2	2	src/modules/mpqs.c

simplify mpqs_eval_cand: remove impossible case

9	13	src/modules/mpqs.c

cleanup mpqs_poly_ctor

10	22	src/modules/mpqs.c

add missing dependence on mpqs.h in Makefile:mpqs.c + fix comment

4	0	config/Makefile.SH
1	1	src/modules/mpqs.h

return mpqs's target_rels

no need to aim for large overshoot: the hashtables kill off all duplicates now

1	1	src/modules/mpqs.c

strtoclosure -> snm_closure [significantly reduce overhead]

1	1	src/basemath/QX_factor.c
4	4	src/basemath/alglin1.c
2	1	src/basemath/alglin2.c
1	1	src/basemath/alglin3.c
3	3	src/basemath/aprcl.c
1	1	src/basemath/elliptic.c
3	3	src/basemath/lfunutils.c
5	3	src/basemath/polarit3.c
4	2	src/basemath/polmodular.c
3	2	src/language/es.c

revert 2cb38ca0, 11- strtoclosure, strtofunction are now shallow

0	3	CHANGES
0	3	COMPAT
4	3	doc/usersch5.tex
2	1	src/language/compile.c

fix polzagier when result is 0

+ fix inefficiency in RgX_Rg_div[exact] in this case

6	4	src/basemath/RgX.c
3	1	src/language/sumiter.c
2	1	src/test/32/pol
2	0	src/test/in/pol

typo

1	1	doc/develop.tex

11- strtoclosure, strtofunction are now shallow

Fixes large overhead and increased memory usage due to argument copies.
Use gerepilecopy or gcopy in the caller if you need garbage collection
or deep copies.

3	0	CHANGES
3	0	COMPAT
3	5	doc/usersch5.tex
6	14	src/language/compile.c

clarify mt_queue_submit

13	8	doc/develop.tex

cleanup RgX_gcd

3	3	src/basemath/polarit2.c

division by t_POLMOD was often wasteful (compute inverse once)

16	3	src/basemath/RgX.c
12	2	src/basemath/polarit2.c

New libpari function QXQX_gcd

1	1	CHANGES
5	0	doc/usersch5.tex
9	0	src/basemath/nffactor.c
11	0	src/basemath/polarit2.c
1	0	src/headers/paridecl.h

New libpari function ZXQX_gcd

1	1	CHANGES
3	0	doc/usersch5.tex
5	1	src/basemath/nffactor.c
1	1	src/basemath/polarit2.c
1	0	src/headers/paridecl.h

14- nfgcd wouldn't terminate [#2159]

1	0	CHANGES
1	1	src/basemath/nffactor.c
17	1	src/test/32/gcdext
4	0	src/test/in/gcdext

13- quadtofp(quadgen(-3)) -> t_FRAC component [#2158]

1	0	CHANGES
10	13	src/basemath/gen2.c
3	3	src/test/32/div

as follow -> as follows

1	2	doc/usersch5.tex
1	1	src/functions/number_fields/idealramgroups

cleanup diffop doc

22	19	src/functions/polynomials/diffop

rewrite develop.tex's Chapter 3 and include the pari-mt.c example

122	66	doc/develop.tex
21	39	examples/pari-mt.c

Add RgM_RgC_mul_fast

49	6	src/basemath/RgV.c

new libpari function FqC_to_mod

1	1	CHANGES
2	0	doc/usersch5.tex
15	0	src/basemath/FpV.c
1	0	src/headers/paridecl.h

new libpari function QM_sqr

1	0	CHANGES
2	0	doc/usersch5.tex
1	1	src/basemath/RgV.c
10	0	src/basemath/ZV.c
1	0	src/headers/paridecl.h

new libpari function FlxV_Flv_multieval

1	1	CHANGES
4	0	doc/usersch5.tex
13	0	src/basemath/Flx.c
1	0	src/headers/paridecl.h

New libpari function F2v_and_inplace

1	1	CHANGES
4	0	doc/usersch5.tex
19	0	src/basemath/F2v.c
1	0	src/headers/paridecl.h

improve parallel.tex

1	1	doc/appa.tex
138	119	doc/parallel.tex

mfcuspval(mf,f,1/5) -> division by 0

mf=mfinit([470,2],0);f=mfeigenbasis(mf)[7];

If we write f = F/E (eisenspaceinit), take into account that we may have
valuation(E | ga)> 0

3	1	CHANGES
16	6	src/basemath/mftrace.c

whitespace edits

41	41	src/modules/mpqs.h

remove useless typedef mpqs_uint64_t

0	5	src/modules/mpqs.h

fix comment

1	3	src/modules/mpqs.h

MPQS_DEBUG_AVMA -> MPQS_DEBUG

4	4	src/modules/mpqs.c

cleanup mpqs_sieve

19	24	src/modules/mpqs.c

mpqs: remove initial warning "this factorization will take several hours"

2	6	src/modules/mpqs.c

remove MPQS_DEBUG_VERYVERBOSE

0	3	src/modules/mpqs.c
0	6	src/modules/mpqs.h