propagate the variable domain to the assumptions database

[rws]

While this works as expected:

sage: n = var('n')
sage: assume(n, 'integer')
sage: solve([n^2 == 3],n)
[]

this does not:

sage: forget()
sage: n = var('n', domain='integer')
sage: solve([n^2 == 3],n)
[n == -sqrt(3), n == sqrt(3)]

So, either solve should also look at the variable domain, or better, let's propagate domain settings to the assumption list. Also:

sage: _ = var('n', domain='integer')
sage: n.is_integer()
False

Creating a non-complex var will invoke Maxima with this. I think this is acceptable for now. Alternatively, examine if assumptions can be done lazily, i.e., before any call to Maxima.

[rws]

New commits:

​e7fe81a

18695: propagate the variable domain to the assumptions database

[git]

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

​95f63a6

18695: fix first attempt; enable exception for wrong domain

[git]

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

​e103b56

18695: fix indentation

[rws]

Together with #18877 this should be able to doctest:

sage: _ = var('n', domain='integer')
sage: n.is_integer()
True

[rws]

Replying to rws:

Together with #18877 this should be able to doctest:

sage: _ = var('n', domain='integer')
sage: n.is_integer()
True

No, this would actually need #18921.

[vdelecroix]

Hello,

It looks good. Though I found confusing that the functions sage_domain_to_ginac and sage_domain_to_maxima have the same name. The first one return something (and does nothing) while the second one is communicating with maxima.

[git]

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

​d3bf439	Merge branch 'develop' into t/18695/setting_domain_integer_on_var_has_no_effect_with_solve
​c142a5f	18695: clearer names for helpers