Changes commited to PARI GIT during the last two weeks

refactor and simplify blocksearch + search clone

+ fix comments

28	30	src/language/init.c

failed to remove one QSCOMP typecast

1	1	src/basemath/FpE.c

better document 'blocks'

8	6	doc/usersch5.tex

remove pdswap

0	1	src/headers/paripriv.h

umael2 -> umael

1	1	src/basemath/hgm.c

paripriv.h / qsort: remove the typecast-to-QSCOMP hack

5	1	src/basemath/FpE.c
0	3	src/headers/paripriv.h
3	2	src/language/gplib.c
3	2	src/test/tune.c

gphelp.in: use single-quoted here-docs [#2621] (Emmanuel Thomé)

Configure replaces @version@ or @datadir@ by some (possibly weird)
character string. The resulting gphelp script assigns this to a perl variable.
Make sure no expansion takes place at this point by using here-docs.

12	3	doc/gphelp.in

32- export and document dblllog2, dbllambertW0, dbllambertW_1

1	0	CHANGES
19	3	doc/usersch5.tex
3	0	src/headers/paridecl.h
0	3	src/headers/paripriv.h

AP  31- [libpari] alg_get_invol

1	0	CHANGES
2	0	doc/usersch6.tex

30- [libpari] document char_renormalize

1	0	CHANGES
16	5	doc/usersch6.tex

29- [libpari] bnrconductor_factored_arch, bnrtoprimitive

Rewrite bnr_*_sanitize in terms of the new primitives.

Rewrite the old bnrconductor variants in terms of the new primitive
bnrconductor_factored_arch or bnrtoprimitive. Then a map can be created
using bnrsurjection and all appropriate objects mapped from the original
bnr representation to the primitive one.

The old (not properly documented) bnrconductormod is now static and
unused outside of the outdated bnrconductor and rnfconductor interface.
(Replaced by bnr_*_sanitize or low level bnrprimitive + bnrsurjection +
abmap_XXX functions)

3	0	CHANGES
2	0	COMPAT
49	10	doc/usersch6.tex
120	120	src/basemath/buch3.c
8	2	src/basemath/kummer.c
1	3	src/functions/number_fields/bnrconductor
2	2	src/headers/paridecl.h
1	0	src/headers/pariold.h

New libpari function FpM_trace

1	1	CHANGES
2	0	doc/usersch5.tex
7	0	src/basemath/alglin2.c
1	0	src/headers/paridecl.h
0	11	src/modules/algebras.c

New libpari function ZM_trace

1	1	CHANGES
2	0	doc/usersch5.tex
12	0	src/basemath/alglin2.c
1	0	src/headers/paridecl.h
0	10	src/modules/algebras.c

New libpari function ZMV_producttree

1	0	CHANGES
9	0	doc/usersch5.tex
25	0	src/basemath/arith1.c
1	0	src/headers/paridecl.h

9- rnfconductor: return a bnr not including generators

This is the default for bnrinit and saves time (no PARI function needs them)

2	0	CHANGES
2	0	COMPAT
1	1	src/basemath/buch3.c
2	3	src/test/32/rnf

charker: reduce result mod full group structure (not only exponent)

+ avoid now useless hnfmodid in bnr_subgroup_sanitize

12	3	src/basemath/buch3.c
1	1	src/basemath/char.c

simplify bnrconductor_factored_i usage

+ document proper use in comments
+ FIXME: bnrconductor_factored and bnrconductor_raw are probably not the
  best way to use bnrconductor_factored_i; we could probably drop them for
  a more flexible interface.
+ FIXME: remove bnrconductormod
+ TODO: the format [idealfactor(f0), foo] or [f0, foo] for idealstar and
  derived functions is not great. It forces either a factorization (bad) or
  a factorback (useless when f0 is known as an ideal, which is common). For
  instance we could allow [F0, foo] where F0 is either idealfactor(f0), f0,
  or [f0, idealfactor(f0)]
Possible use case: same f0, variable foo; for instance when the
conductor at infinity is smaller than the modulus but the finite
conductor is spot on, and we need to sanitize the bnr by one attacked to
the conductor.

17	16	src/basemath/buch3.c

rewrite bnrconductormod in terms of bnrconductor_factored_i

34	44	src/basemath/buch3.c

24- lfun(bnr, vector of chars) -> impossible division [#2623]

The k characters have the same conductor F. The error could happen when they
had different orders [o1, ..., ok] and we initially computed
the ray class group bnr mod F, modulo o1-th powers. Must be done modulo
o-th powers, where o is the lcm(o1,...,ok).

1	0	CHANGES
52	4	src/basemath/buch3.c
3	31	src/basemath/lfunutils.c
2	0	src/headers/paridecl.h
4	2	src/test/32/lfun
4	0	src/test/in/lfun

cleanup further bnrconductormod + bnrconductor_factored_i

6	6	src/basemath/buch3.c

cleanup bnrconductormod + bnrconductor_factored_i

14	10	src/basemath/buch3.c

bnrconductormod: don't shadow 'bid'

2	2	src/basemath/buch3.c

whitespace edit

1	1	src/basemath/buch3.c

fix comment

3	6	src/basemath/buch3.c

ComputeKernel: use map as argument instead of two bnr

7	7	src/modules/stark.c

actually remove bnrchar_primitive_raw

0	4	src/basemath/buch3.c

10- remove bnrchar_primitive, bnrchar_primitive_raw.

Use bnrmap. Most of the time, the map can be precomputed: wasteful to do
it for each chi.

2	0	CHANGES
2	0	COMPAT
0	13	doc/usersch6.tex
10	18	src/basemath/buch3.c
0	2	src/headers/paridecl.h
4	3	src/modules/stark.c

move char_denormalize to char.c

0	16	src/basemath/buch3.c
17	0	src/basemath/char.c

export abmap_char_image, abmap_nchar_image

1	0	CHANGES
10	2	doc/usersch6.tex
2	2	src/basemath/buch3.c
8	7	src/basemath/lfunutils.c
2	0	src/headers/paridecl.h

lfunutils: minor gc cleanups

6	11	src/basemath/lfunutils.c

Flx_genus2trace_naive: unroll Flv_diffnext

7	9	src/basemath/lfunutils.c

New libpari function zero_FlxM_copy

1	0	CHANGES
2	0	doc/usersch5.tex
9	0	src/basemath/Flx.c
1	0	src/headers/paridecl.h