lazy import attributes of a class are not substituted back after the import

[nthiery]

sage: class A:
....:     Associative = LazyImport('sage.categories.magmas', 'Magmas')
....: 
sage: A.Associative
<class 'sage.categories.magmas.Magmas'>
sage: type(A.__dict__["Associative"])
<type 'sage.misc.lazy_import.LazyImport'>

Component: misc

Issue created by migration from https://trac.sagemath.org/ticket/15648

[simon-king-jena]

comment:1

See my comment at #10963. We could simply create a lazy class attribute with an import statement inside. Proof of concept:

sage: def imported_lazy_class_attribute(module_name, cls_name):                         
....:     return lazy_class_attribute(lambda cls: getattr(__import__(module_name, {}, {}, [cls_name]),cls_name))
....: 
sage: class Test(object):
....:     Finite = imported_lazy_class_attribute('sage.categories.finite_permutation_groups', 'FinitePermutationGroups')
....:     
sage: Test.Finite
<class 'sage.categories.finite_permutation_groups.FinitePermutationGroups'>

Alternatively, since LazyImport has some __get__: We could make the lazy import behave as a class attribute, if it is used as attribute of a class (__get__ knows whether it is used as attribute of a class.

[simon-king-jena]

comment:2

Strange. Shouldn't LazyImport already act as a class attribute? After all, I see the following code:

            if owner is None:
                if self._namespace and self._namespace[alias] is self:
                    self._namespace[alias] = self._object
            else:
                from inspect import getmro
                for cls in getmro(owner):
                    if cls.__dict__.get(alias, None) is self:
                        setattr(cls, alias, self._object)
                        break

This should actually do the trick.

[simon-king-jena]

comment:3

In the example of the ticket description, I suppose you simply need to proved as_name="Associative", and then it should work. Can't test it right now, though.

[simon-king-jena]

comment:4

The only (mild) problems I see:

1. Why is the _get_object, if it does what ostensibly is the job of __get__?
2. Why is the imported object put into the dict of a class, but not into the dict of an instance? _get_object can't do it, since it doesn't know the instance.

Hence, I am all for removing _get_object.

[simon-king-jena]

Work Issues: Do we need a lazy import to act as lazy instance attribute?

[simon-king-jena]

comment:5

sage: from sage.misc.lazy_import import LazyImport
sage: class A:
....:     Associative = LazyImport('sage.categories.magmas', 'Magmas', 'Associative')
....:     
sage: A.Associative
<class 'sage.categories.magmas.Magmas'>
sage: A.__dict__['Associative']
<class 'sage.categories.magmas.Magmas'>

Hence, unless you agree that we should change the lazy import statement so that it additionally acts as a lazy instance attribute, then this ticket is invalid.

[jdemeyer]

Author: Jeroen Demeyer

[jdemeyer]

Dependencies: #22752

[jdemeyer]

Changed work issues from Do we need a lazy import to act as lazy instance attribute? to none

[jdemeyer]

Branch: u/jdemeyer/ticket/15648

[jdemeyer]

Commit: d5f2348

[jdemeyer]

New commits:

94302ca	Do not declare functions/methods as "cdef inline"
8aac18a	Various improvements to lazy imports
8ec9f5d	Lazy import context based on __import__
d50f9a6	Make "with lazyimport" context more thread-safe
3d62c9e	Update documentation
d5f2348	Search for object in __get__

[sagetrac-git]

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

9e1208b	lazyimport -> _lazyimport_
796a452	Search for object in __get__

[sagetrac-git]

Changed commit from d5f2348 to 796a452

[jdemeyer]

Changed commit from 796a452 to none

[jdemeyer]

Changed author from Jeroen Demeyer to none

[jdemeyer]

Changed branch from u/jdemeyer/ticket/15648 to none

[jdemeyer]

Changed dependencies from #22752 to none