enhance GroupLibGAP

[Vincent Delecroix]

• fix libgap conversion of elements
• faster group operations
• more methods

[Vincent Delecroix]

New commits:

​d90c228	fix libgap conversion of ElementLibGAP
​d25f239	allow raag elements to be constructed from libgap object
​03dca2b	rework gens of ParentLibGAP
​2044eb2	less indirection in group operations for ElementLibGAP
​95c76cb	clean import
​81e7ecf	more capable libgap mixin and libgap group

[Simon Brandhorst]

the part on class Element(ArtinGroupElement): looks unrelated. Please move that to a different ticket. Otherwise it looks good to me.

[Vincent Delecroix]

Replying to sbrandhorst:

the part on class Element(ArtinGroupElement): looks unrelated. Please move that to a different ticket. Otherwise it looks good to me.

It is related! Without this, the ArtinGroupElement can not be rebuilt from libgap objects. Without this change, the two tests introduced in commit d90c228 just fail.

[Simon Brandhorst]

Whether or not the internal logic is okay, I do not know. The code seems to do what it says. So I am okay.

[Vincent Delecroix]

Thank you!

[Volker Braun]

On Python 2:

**********************************************************************
File "src/sage/groups/libgap_wrapper.pyx", line 505, in sage.groups.libgap_wrapper.ElementLibGAP.gap
Failed example:
    type(libgap(FreeGroup('a, b').an_element()))
Expected:
    <class 'sage.libs.gap.element.GapElement'>
Got:
    <type 'sage.libs.gap.element.GapElement'>
**********************************************************************
1 item had failures:
   1 of   9 in sage.groups.libgap_wrapper.ElementLibGAP.gap
    [141 tests, 1 failure, 0.59 s]
----------------------------------------------------------------------
sage -t --long src/sage/groups/libgap_wrapper.pyx  # 1 doctest failed
----------------------------------------------------------------------

[Vincent Delecroix]

Is this supposed to fail? There are tons of doctest with randomly chosen <class X> or <type X>!

$ git grep "<class " | wc -l
1627
$ git grep "<type " | wc -l
937

[Travis Scrimshaw]

Replying to vdelecroix:

Is this supposed to fail? There are tons of doctest with randomly chosen <class X> or <type X>!

It comes from a difference between Python2 and 3, and they are not randomly chosen. However, you can use <... X> to make it compatible with both.

[Vincent Delecroix]

sage/groups/abelian_gps/abelian_group.py line 961-962

sage: F = AbelianGroup(3, [2], names='abc')
sage: list(map(type, F.gens_orders()))
[<type 'sage.rings.integer.Integer'>,
 <type 'sage.rings.integer.Integer'>,
 <type 'sage.rings.integer.Integer'>]

Which is not the output you get with Python 3. Should I just change it to type then (it makes sage -t happy).

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​74d7935

doctest compatibility with Python 2