Changes commited to PARI GIT during the last two weeks

5- ellisdivisible: potential oo-loop [#2695]

1	0	CHANGES
1	1	src/basemath/elliptic.c
6	0	src/test/in/ell

fix setprec to round up as nbits2lg would

don't assume "precision" argument is a multiple of BITS_IN_LONG

1	1	src/headers/parigen.h

improve doc of mk*n functions + don't document mkintn in chapter 4

6	19	doc/usersch4.tex
11	5	doc/usersch5.tex

doc_make: use void for no args

1	0	src/desc/doc_make

O: only keep Doc and Help so that ??O work

2	3	src/functions/polynomials/O

phi5_ZV: mkintn arguments are unsigned ints, not ulongs [#2692]

Created problems on the emscripten platform

13	13	src/basemath/polmodular.c

clarify doc

8	6	doc/usersch5.tex

6- readline completion: don't include yet undefined symbols [#2694]

If identifier valence is EpNEW, skip it in  completion.

Rationale: identifier exists in parser hashtables but it has no
definition yet. It is likely that it was created by mistake (e.g., typo)
or that it has been kill-ed by the user. In any case, its value is NULL.
It makes sense not to propose it in completions in interactive use until
it acquires an actual value.

1	0	CHANGES
1	1	src/language/readline.c

cleanup mflinear and simplify mf2linear

26	22	src/basemath/mftrace.c

mflinear(F,L): add sanity checks for L

When coefficients t_POLMOD in the t variable, validate that they are
defined modulo charpoly(chi)

12	2	src/basemath/mftrace.c
7	1	src/test/32/mf
2	0	src/test/in/mf

simplify mf2linear

9	15	src/basemath/mftrace.c

4- mflinear(MF in half integer weight, coefs in Q(chi)): wrong result

Possible confusion between Q(chi) and Q(chi1) where chi1 is the character
for f * mfTheta, where chi1 = chi or chi * (-4/.). In the latter case
the cyclotomic fields are isomorphic, but only identical when the order of
chi is 0 mod 4.

mf=mfinit([28,5/2,Mod(4,7)],1);y=Mod([t,t,t,t],t^2+t+1);
F1=mflinear(mf,y);
F2=mflinear(mfbasis(mf),y);
mfcoefs(F1,10)
mfcoefs(F2,10)

1	0	CHANGES
61	7	src/basemath/mftrace.c
32	1	src/test/32/mf
9	0	src/test/in/mf

typo in previous commit f8a9ba81fb0: H instead of gel(H,1)

6	3	src/basemath/subfield.c

subfields: check group type before using group_order [#2691]

galoissubgroups can now return subgroups in "list of elements" format
(non-WSS case). Which group_order doest not support

5	3	src/basemath/subfield.c

remove isolated checks for prec < LOWDEFAULTPREC

0	1	src/basemath/gen2.c
0	2	src/basemath/rootpol.c
2	12	src/basemath/trans1.c
0	1	src/basemath/trans2.c

typo in compatibility #define real_1_bit

1	1	src/headers/pariold.h

Zp_divlift: handle correctly b = NULL when n = 1

followup to bf8c1198f3a

2	1	src/basemath/Zp.c

fix RgX_galconj_bound + use vandermondeinverseinit_real

incorrectly called QX_complex_roots on a t_POL with t_REAL coefficients.

30	8	src/basemath/nffactor.c
1	0	src/headers/paripriv.h

vandermondeinverseinit_real: complex interpolation up to conjugation

Allows to reconstruct and algebraic number given its r1 + r2 non-conjugate
complex embeddings (instead of including the redundant conjugate ones).

Stabler and saves a factor (r1+r2) / (r1+2r2) (so nothing for totally
real fields and 2 for totally complex).

Utility function ZX_nf_roots returning complex roots of a ZX up to conjugation
(as nf=nfinit(T); nf.roots)

Use in galoisborne and applications

2	1	CHANGES
29	11	doc/usersch5.tex
5	0	doc/usersch6.tex
21	17	src/basemath/base1.c
52	7	src/basemath/bibli2.c
37	17	src/basemath/galconj.c
21	8	src/basemath/nffactor.c
2	8	src/basemath/subfield.c
2	0	src/headers/paridecl.h
1	1	src/headers/paripriv.h

Zp_divlift: handle n=1 correctly

1	1	src/basemath/Zp.c

bound_for_coeff: wrong computation of Mahler measure

confusion between nf.roots (modulo complex conjugation) and initgaloisborne
(all complex roots).

+ avoid squares computations

47	19	src/basemath/subfield.c