Changes commited to PARI GIT during the last two weeks

rnfelttrace, rnfeltnorm: wrong interpretation of input [#2509]

15- rnfelttrace(rnf, t_POL in rnf.pol variable): wrong
16- rnfeltnorm(rnf, t_POL in rnf.pol variable): wrong

2	0	CHANGES
5	0	src/basemath/base3.c
13	21	src/test/32/rnf

use gc_const

1	1	src/basemath/base5.c

14- ellsaturation was leaking variable numbers

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

fix docs

0	1	src/functions/number_fields/rnfeltnorm
0	1	src/functions/number_fields/rnfelttrace

fix doc

4	3	src/functions/number_fields/HEADER

improve uprimepi

15	15	src/basemath/prime.c

fix comment

1	1	src/basemath/prime.c

11- [libpari] Z_to_famat, Q_to_famat

1	0	CHANGES
10	1	doc/usersch6.tex

13- idealstar(nf,id,,MOD). Exceptions may occur in ideallog [#2549]

In particular when the "generator" for a cyclic F_q^* (\otimes Z/MOD)
doesn't have order q-1 but some strict divisor. In this case, the first
stage of the computation of the discrete log (reducing to input = 1 (mod I))
may fail and we get a meaningless err_COPRIME exception.

Solution: store MOD in the structure in idealstar, and use it in ideallog

3	0	CHANGES
8	1	doc/usersch6.tex
5	2	src/basemath/base3.c
1	0	src/headers/paridecl.h
2	0	src/headers/pariinl.h
2	1	src/test/32/ideal
2	0	src/test/in/ideal

EvalCoeff: remove test for impossible condition

1	3	src/modules/stark.c

GetValue1: don't use l for an 'lg - 1'

3	3	src/modules/stark.c

Fix handling of vcfile. [#2548] from Denis Excoffier.

1	2	config/Makefile.SH

Stark / get_C: save one sqrt + avoid double rounding -> get_P

no effect in practice, but slightly stabler and more efficient

10	10	src/modules/stark.c

CharNewPrec: rewrite so as not to destroy bnr components

5	7	src/modules/stark.c

mapapply: doc: fix typo

2	1	src/functions/programming/mapapply

avoid silly subFB by always using bad_subFB() [#2542]

bad_subFB was only used during initial subFB setup, not when changing it
Use it to avoid ridiculous case like

bnfinit(x^4 - 330*x^2 + 5445)

where many small primes are divisible by a unique (ramified) maximal.
Those are terrible choices for subFB: for instance, if P^e = (p)
[usually e = 2], then taking P^k in rnd_rel produces (up to content) a
result depending on k % e only.

2	1	src/basemath/buch2.c
2	1	src/test/32/bnf
2	0	src/test/in/bnf

nflist: use MEDDEFAULTPREC by default rather than DEFAULTPREC

10	10	src/basemath/nflist.c

galoisinit: doc: fix doc of 3x3:4 case

2	3	src/functions/number_fields/galoisinit

ellheegner: increase precision margin [#2546]

1	1	src/basemath/ellanal.c

Fincke_Pohst_ideal: add back maxtry_ELEMENT

In some case Fincke_Pohst_ideal could loop without factoring elements.

2	0	src/basemath/buch2.c

ZM_lll_norms: Do not compute rank when LLL_NOFLATTER is set

2	2	src/basemath/lll.c

gen_comb: disable flatter.

gen_comb generates high-dimension matrix of low rank for which flatter has no chance.

1	1	src/basemath/qfisom.c