[with patch, positive review] Parent: fixes broken (implicit) invariant between ._element_constructor and self._element_init_pass_parent
In Parent, there is a (previously implicit) invariant, namely that:
self._element_init_pass_parent == guess_pass_parent(between ._element_constructor)
This invariant was broken when self._element_constructor was set from
self._element_constructor_ by call. This made the parent not to be
passed properly to _element_constructor.
This patch fixes this (one line), and adds a corresponding doctest.
It also includes a related trivial fix to a line that is apparently never used in coerce_maps, and adds a comment about this.
The category patch #5891 depends heavily on this one, as this way of
setting _element_constructor becomes the default one for simple
parents.
Change History (11)
| Description: |
modified (diff)
|
| Summary: |
Parent: fixes broken (implicit) invariant between ._element_constructor and self._element_init_pass_parent →
[with patch, needs review] Parent: fixes broken (implicit) invariant between ._element_constructor and self._element_init_pass_parent
|
| Description: |
modified (diff)
|
| Summary: |
[with patch, needs review] Parent: fixes broken (implicit) invariant between ._element_constructor and self._element_init_pass_parent →
[with patch, positive review] Parent: fixes broken (implicit) invariant between ._element_constructor and self._element_init_pass_parent
|
| Resolution: |
→ fixed
|
| Status: |
assigned →
closed
|
| Authors: |
→ Nicolas Thiery
|
| Merged in: |
→ 4.0.1.alpha0
|
| Reviewers: |
→ Robert Bradshaw
|
| Description: |
modified (diff)
|
| Report Upstream: |
→ N/A
|
| Authors: |
Nicolas Thiery →
Nicolas M. Thiéry
|
I should mention that sage -testall passes smoothly with sage-3.4.2-alpha0 (except for a trivial broken test in interfaces.r which also fails before applying the patch). Haven't tried it with sage-3.4.2 final (under compilation)