Changes commited to PARI GIT during the last two weeks

Doc: polclass: missing leading space

1	1	src/functions/polynomials/polclass

bnfinit: improve rnd_rel (missing chunk 2)

1	1	src/basemath/buch2.c

bnfinit: improve rnd_rel (missing chunk)

7	7	src/basemath/buch2.c

document bnr_sanitize

+ use one more time

1	0	CHANGES
5	0	doc/usersch6.tex
1	1	src/basemath/buch3.c

refactor (and simplify) a bnr_sanitize function

12	4	src/basemath/buch3.c
1	2	src/basemath/kummer.c
1	0	src/headers/paridecl.h

bnrclassfieldvec: missing check for variable priority [#2658]

9	3	src/basemath/kummer.c
6	3	src/test/32/bnrclassfield
3	1	src/test/in/bnrclassfield

bnr_vecchar_sanitize: remove useless and misleading bnrc = bnr initialization

bnrc is recomputed anyway later. And it better be since bnr could be a
yet unsanitized bnf.

1	1	src/basemath/buch3.c

fix comment

2	2	src/basemath/ifactor1.c

bnfinit: improve rnd_rel by allowing to change test modulus

47	34	src/basemath/buch2.c
1	1	src/functions/number_fields/bnfinit
1	1	src/headers/paripriv.h
2	2	src/test/32/bnf
1	1	src/test/in/bnr

quadunitnorm: avoid full factorization of D if possible

84	0	src/basemath/ifactor1.c
2	14	src/basemath/quad.c
1	0	src/headers/paridecl.h

improve comments

2	1	src/basemath/ifactor1.c

moebius, ispowerful: avoid duplicate trial division by 3,5,7

3	3	src/basemath/ifactor1.c

quadunitnorm: test for prime divisor = 3 mod 4 [#2657]

Then norm is +1. We can factor D (at a subexponential cost): the naive
continued fraction expansion is much more costly, in O~(D^(1/2))

14	0	src/basemath/quad.c
2	1	src/test/32/quadclassunit
1	0	src/test/in/quadclassunit

bnf_increase_LIMC(LIMC, D): allow answer> D

Primes of norm <= D generate Cl(K) under GRH. Nevertheless, the index
calculus algorithm may fail when the bound is too small (for instance
because the sub factorbase for random relations has a single element).
Useful for small discriminints, e.g.,

setrand(40567);bnf=bnfinit(x^2-x-1)

(oo loop)

12	10	src/basemath/buch1.c
3	0	src/test/in/bnf

subFB_change: don't shadow the 'l' variable in internal loops

Functionally equivalent to previous code

2	6	src/basemath/buch2.c

fix documentation for factor over Q(i) [#2654]

factors are ordered by increasing imaginary part

5	4	src/functions/number_theoretical/factor

fix comment

1	1	src/basemath/buch1.c