random_element methods should all accept *args and **kwds

[malb]

Because:

• some more highlevel random_element functions pass thru *args and **kwds
• it unifies the interface to some extend
• it allows the user to pass thru parameters

Apply

• patches from #8955
• trac_3893_args_kwds_for_random_element_4.6.alpha2.patch​

[niles]

I've reinterpreted this ticket as "random_element methods which call other random_element methods should pass on *args and kwds". This applies mostly to sage.rings, but also a few other cases. The list below shows the output of sage: search_src("def random_element"), annotated by the changes this patch makes.

make ptestlong passes all tests, and documentation builds without errors or warnings. Details below.

Note that #9481 fixes random_element for univariate power series rings, and I've ignored padics since there is a large revision in process: e.g. #6084 and #5075.

• algebras/quatalg/quaternion_algebra.py:448: def random_element(self, *args, kwds):
  • nothing to fix

• algebras/quatalg/quaternion_algebra.py:1142: def random_element(self, *args, kwds):
  • nothing to fix

• categories/enumerated_sets.py:472: def random_element(self):
  • not implemented

• categories/infinite_enumerated_sets.py:83: def random_element(self):
  • not implemented

• coding/linear_code.py:2022: def random_element(self):
  • FIXED

• combinat/cartesian_product.py:177: def random_element(self):
  • doesn't call other random_element()

• combinat/choose_nk.py:148: def random_element(self):
  • doesn't call other random_element()

• combinat/multichoose_nk.py:90: def random_element(self):
  • doesn't call other random_element()

• combinat/partition.py:3269: def random_element(self, measure = 'uniform'):
  • doesn't call other random_element()

• combinat/partition.py:3297: def random_element_uniform(self):
  • doesn't call other random_element()

• combinat/partition.py:3360: def random_element_plancherel(self):
  • doesn't call other random_element()

• combinat/permutation.py:2792: def random_element(self):
  • doesn't call other random_element()

• combinat/permutation.py:2991: def random_element(self):
  • doesn't call other random_element()

• combinat/permutation.py:3110: def random_element(self):
  • doesn't call other random_element()

• combinat/permutation.py:3334: def random_element(self):
  • doesn't call other random_element()

• combinat/permutation_nk.py:112: def random_element(self):
  • doesn't call other random_element()

• combinat/split_nk.py:113: def random_element(self):
  • doesn't call other random_element()

• combinat/subset.py:238: def random_element(self):
  • doesn't call other random_element()

• combinat/subset.py:485: def random_element(self):
  • calls random_element() of ChooseNK from sage.combinat.choose_nk, and that method takes no arguments

• combinat/tableau.py:2162: def random_element(self):
  • doesn't call other random_element()

• crypto/mq/mpolynomialsystemgenerator.py:184: def random_element(self):
  • not implemented

• crypto/mq/sr.py:1089: def random_element(self, elem_type = "vector"):
  • FIXED

• groups/group.pyx:196: def random_element(self, bound=None):
  • not implemented

• groups/abelian_gps/abelian_group.py:854: def random_element(self):
  • doesn't call other random_element()

• groups/abelian_gps/abelian_group_element.py:312: def random_element(self):
  • doesn't call other random_element()

• groups/abelian_gps/dual_abelian_group.py:221: def random_element(self):
  • doesn't call other random_element()

• groups/matrix_gps/matrix_group.py:643: def random_element(self):
  • doesn't call other random_element()

• groups/perm_gps/permgroup.py:1107: def random_element(self):
  • doesn't call other random_element()

• matrix/matrix_space.py:1221: def random_element(self, density=None, *args, kwds):
  • nothing to fix

• modular/dirichlet.py:2290: def random_element(self):
  • doesn't call other random_element()

• modular/arithgroup/congroup_sl2z.py:142: def random_element(self, bound=100):
  • nothing to fix

• DONE: modules/free_module.py:1710: def random_element(self, prob=1.0, *args, kwds):
  • FIXED

• modules/free_module.py:4147: def random_element(self, prob=1.0, *args, kwds):
  • FIXED

• modules/fg_pid/fgp_module.py:1307: def random_element(self, *args, kwds):
  • FIXED

• rings/complex_double.pyx:214: def random_element(self, double xmin=-1, double xmax=1, double ymin=-1, double ymax=1):
  • doesn't call any other random_element() method

• rings/complex_field.py:421: def random_element(self, component_max=1):
  • FIXED

• rings/complex_interval_field.py:334: def random_element(self, *args):
  • FIXED

• rings/contfrac.py:331: def random_element(self, num_bound=2, den_bound=2):
  • FIXED

• rings/fraction_field.py:573: def random_element(self, *args, kwds):
  • nothing to fix

• rings/integer_ring.pyx:511: def random_element(self, x=None, y=None, distribution=None):
  • doesn't call any other random_element() method

• rings/pari_ring.py:58: def random_element(self, bound=0):
  • always returns 0

• rings/power_series_ring.py:622: def random_element(self, prec=None, *args, kwds):
  • fixed by #9481

• rings/rational_field.py:762: def random_element(self, num_bound=None, den_bound=None, distribution=None):
  • FIXED

• rings/real_double.pyx:406: def random_element(self, double min=-1, double max=1):
  • doesn't call any other random_element() method

• rings/ring.pyx:865: def random_element(self, bound=2):
  • uses python's randint instead of ZZ.random_element()
  • (this has the disadvantage of not allowing one to specify the distribution, but probably no one cares)

• rings/real_mpfi.pyx:666: def random_element(self, *args):
  • FIXED

• rings/real_mpfr.pyx:884: def random_element(self, min=-1, max=1, distribution=None):
  • doesn't call any other random_element() method

• rings/finite_rings/element_givaro.pyx:352: def random_element(self, *args, kwds):
  • nothing to fix

• rings/finite_rings/finite_field_base.pyx:569: def random_element(self, bound=None):
  • FIXED

• rings/finite_rings/finite_field_givaro.py:180: def random_element(self, *args, kwds):
  • nothing to fix

• rings/finite_rings/integer_mod_ring.py:1061: def random_element(self, bound=None):
  • calls sage.rings.CommutativeRing?.random_element(), whose only allowed argument is 'bound'

• rings/number_field/number_field.py:1247: def random_element(self, num_bound=None, den_bound=None,
  • calls _random_element() method of sage.rings.number_field.number_field_element.NumberFieldElement_absolute, and already includes all possible keyword arguments

• rings/number_field/number_field_ideal.py:1132: def random_element(self, *args, kwds):
  • nothing to fix

• rings/number_field/order.py:858: def random_element(self, *args, kwds):
  • nothing to fix

• rings/padics/generic_nodes.py:265: def random_element(self, algorithm='default'):
  • not touching padics (see above)

• rings/padics/padic_base_leaves.py:508: def random_element(self, algorithm='default'):
  • not touching padics (see above)

• rings/padics/padic_extension_generic.py:286: def random_element(self):
  • not touching padics (see above)

• rings/polynomial/laurent_polynomial_ring.py:707: def random_element(self, low_degree = -2, high_degree = 2, terms = 5, choose_degree=False,*args, kwds):
  • nothing to fix

• rings/polynomial/multi_polynomial_ring_generic.pyx:585: def random_element(self, degree=2, terms=None, choose_degree=False,*args, kwargs):
  • (added by #1956) nothing to fix

• rings/polynomial/pbori.pyx:899: def random_element(self, degree=None, terms=None, choose_degree=True, vars_set=None):
  • doesn't call any other random_element() method

• rings/polynomial/polynomial_quotient_ring.py:591: def random_element(self, *args, kwds):
  • nothing to fix

• rings/polynomial/polynomial_ring.py:883: def random_element(self, degree=2, *args, kwds):
  • nothing to fix

[malb]

I rebased the patch to 4.5.3.alpha2 and ran long doctests: all passed. The patch also looks good, the only issue I'd have: I'd either move the lines "Trac#3893 ..." to the TESTS:: section or remove that line completely. I understand that people want to point out where something changed, but that docstring is saying nothing to a user since it is developer information.

[niles]

Thanks for taking a look!

Somewhere I gained the impression that doctests for a ticket should mention the ticket number--is that not necessarily the case? I don't really have a strong opinion, but since it may be non-obvious functionality I'm in favor of keeping these tests in the EXAMPLES:: section. How about rewording the description to something like the following?

Passes extra positional or keyword arguments through (Trac #3893)::

[malb]

Replying to niles:

Thanks for taking a look! Somewhere I gained the impression that doctests for a ticket should mention the ticket number--is that not necessarily the case? I don't really have a strong opinion, but since it may be non-obvious functionality I'm in favor of keeping these tests in the EXAMPLES:: section. How about rewording the description to something like the following? Passes extra positional or keyword arguments through (Trac #3893)::

I'm not sure what the official policy is, but I'd say it would be a bit odd when in a few years time the docstring is cluttered with trac ticket numbers, what information do they communicate that is relevant? I am very much in favour of having "Passes extra positional or keyword arguments through::". I leave it up to you, so you have a positive review, i.e. toggle it yourself if you don't want to change it.

[niles]

clarified docstrings

[niles]

Replying to malb:

I'm not sure what the official policy is, but I'd say it would be a bit odd when in a few years time the docstring is cluttered with trac ticket numbers

fair enough; I also can't find any mention of a policy like this in the developer guide, so I've changed the docstring as you suggested. Since I don't have 4.5.3a2 installed, will you re-verify that the patch applies cleanly before hitting "positive review"? (Since the changes were easy, I just made them directly to the rebased patch file itself.)

[niles]

rebased to 4.5.3

[niles]

rebased patch applies cleanly to 4.5.3; switching to positive review, as per Martin's recommendation

[mpatel]

With the latest patch applied to a trial 4.6.alpha2, I get two doctest failures:

sage -t -long "devel/sage/sage/rings/contfrac.py"           
**********************************************************************
File "/mnt/usb1/scratch/mpatel/tmp/sage-4.6.alpha1-working/devel/sage/sage/rings/contfrac.py", line 40:
    sage: a
Expected:
    [    [-1, 2] [-1, 1, 94]      [0, 2]       [-12]]
    [       [-1]      [0, 2]  [-1, 1, 3]   [0, 1, 2]]
    [    [-3, 2]   [0, 1, 2]        [-1]         [1]]
    [       [-1]      [0, 3]         [1]        [-1]]
Got:
    [    [-1, 2] [-1, 1, 94]      [0, 2]       [-12]]
    [       [-1]      [0, 2]  [-1, 1, 3]   [0, 1, 2]]
    [    [-3, 2]         [0]   [0, 1, 2]        [-1]]
    [        [1]        [-1]      [0, 3]         [1]]
**********************************************************************
File "/mnt/usb1/scratch/mpatel/tmp/sage-4.6.alpha1-working/devel/sage/sage/rings/contfrac.py", line 46:
    sage: f
Expected:
    [1]*x^4 + [2]*x^3 + ([-12, 1, 14, 7, 1, 1, 7])*x^2 + ([-40, 2, 32, 2, 1, 1, 2])*x + [8, 1, 5, 2, 14, 1, 1, 3]
Got:
    [1]*x^4 + ([-2, 3])*x^3 + [14, 1, 1, 1, 9, 1, 8]*x^2 + ([-13, 4, 1, 2, 1, 1, 1, 1, 1, 2, 2])*x + [-6, 1, 5, 9, 1, 5]
**********************************************************************

Perhaps there's an interaction with one of the ​already merged tickets?

[niles]

addressed interaction with #8955

[niles]

Indeed; the interaction seems to be with #8955: after applying those patches, I get exactly the same doctest failures, with exactly the same "Got" output. I just uploaded a fixed patch which passes all doctests after applying #8955.

[mpatel]

Martin, could you review the latest patch? If it's OK, I'll merge it into 4.6.alpha3.

[malb]

The updated patch applies with some fuzz, but it applies. Doctests pass. I trust only doctest failures were addressed and thus I give this a positive review.