Py3: Fix libs.ntl.ntl_ZZ_pX.pyx doctests.

[vinklein]

Fix libs.ntl.ntl_ZZ_pX.pyx doctests for python3. All libs.ntl doctests should pass with this ticket.

Component: python3

Author: Vincent Klein, Frédéric Chapoton

Branch/Commit: a93f26b

Reviewer: Vincent Klein, Frédéric Chapoton

Issue created by migration from https://trac.sagemath.org/ticket/27588

[vinklein]

comment:1

Looking at #24804 (which is already in merge confict) commits it doesn't appear to be conflict with this ticket.

[vinklein]

Branch: u/vklein/27588

[vinklein]

Commit: 91bac0e

[vinklein]

New commits:

91bac0e

Trac 27588: Py3: Fix libs.ntl.ntl_ZZ_pX.pyx ...

[vinklein]

Author: Vincent Klein

[fchapoton]

Changed branch from u/vklein/27588 to public/ticket/24804

[fchapoton]

Changed commit from 91bac0e to e2e31f3

[fchapoton]

comment:5

maybe we can instead try again my previous proposal in #24804 ?

---

New commits:

9d1c59f	add a simple issequence implementation
1d7066c	add a special case for the most common case of list or tuple
79d5bdf	py3: some care for libs/ntl
62508e3	py3: replace a troublesome map() in ntl_GF2X._sage_
3407ae3	cover more sequence types here by using issequence
e2e31f3	update some defunct code

[fchapoton]

Changed branch from public/ticket/24804 to public/ticket-27588

[fchapoton]

Changed commit from e2e31f3 to bb4cb87

[fchapoton]

comment:6

here is a simple but working solution

---

New commits:

bb4cb87

fix for ntl (simple minded but doing the job)

[sagetrac-git]

Changed commit from bb4cb87 to cd7b2fc

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

cd7b2fc

fix for ntl (simple minded but doing the job)

[fchapoton]

comment:8

should be good now. I also made a few minor changes in the doc.

[fchapoton]

Changed author from Vincent Klein to Vincent Klein, Frédéric Chapoton

[fchapoton]

comment:10

fails on python2

[vinklein]

comment:11

As range is not a class in python2 the new isinstance is supposed to fail and the cythonize seems to be a little inconsistent if you use isinstance with a tuple as a second argument:
elif isinstance(v, (list, tuple, range)):
generate

__pyx_t_1 = PyObject_IsInstance(__pyx_v_v, __pyx_builtin_range); 
__pyx_t_4 = (__pyx_t_1 != 0);

and
elif isinstance(v, (list, tuple)) or isinstance(v, range): generate

pyx_t_4 = PyObject_IsInstance(__pyx_v_v, __pyx_builtin_range); 
if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(0, 96, __pyx_L1_error)

The latest is the consistent behaviour for me.

For the errors generated by this ticket with python2 the object return -1 on PyObject_IsInstance(__pyx_v_v, __pyx_builtin_range) and then isinstance(v, (list, tuple, range) return True.

[fchapoton]

comment:12

peut etre que ca marcherait mieux en utilisant xrange, je n'ai pas le temps..

[vinklein]

comment:13

Je vais essayer.

[vinklein]

comment:14

It looks like that work !

with sage python3:

sage: from past.builtins import xrange
sage: isinstance(range(5), xrange)
True
sage: type(xrange(2))
<class 'range'>

[sagetrac-git]

Changed commit from cd7b2fc to 00f92a6

[sagetrac-git]

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

00f92a6

Trac 27588: use xrange instead of range to be ...

[vinklein]

comment:16

I give positive review to the commits prior to 00f92a6

[fchapoton]

comment:17

probably no need for the import of xrange, as we are in a pyx file..

[sagetrac-git]

Changed commit from 00f92a6 to a93f26b

[sagetrac-git]

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

a93f26b

Trac 27588: remove import xrange

[vinklein]

comment:19

Indeed. import xrange has been removed.

[fchapoton]

comment:20

ok, works both in py2 and py3. If you agree, please set to positive

[fchapoton]

Reviewer: Vincent Klein, Frédéric Chapoton

[vinklein]

comment:21

I agree.

[vbraun]

Changed branch from public/ticket-27588 to a93f26b