AttributeError raised by method __eq__ of poset element

[seblabbe]

Using sage-4.7.2

sage: P = Poset([[1,2],[4],[3],[4],[]])
sage: P(0) == int(0)
Traceback (most recent call last):
...
AttributeError: 'int' object has no attribute 'parent'

This error is raised when ploting a poset :

sage: m = WordMorphism('a->abb,b->ba')
sage: w = m.fixed_point('a')
sage: prefix = Word(list(w[:100]))
sage: pals = prefix.palindromes()
sage: poset = Poset((pals, lambda x,y: x.is_factor(y)))
sage: H = poset.hasse_diagram()
sage: H.plot()
Traceback (most recent call last):
...
AttributeError: 'int' object has no attribute 'parent'

CC: @sagetrac-sage-combinat

Component: combinatorics

Keywords: poset

Author: David Loeffler

Reviewer: Nicolas M. Thiéry, Simon King

Merged: sage-5.0.beta9

Issue created by migration from https://trac.sagemath.org/ticket/12351

[simon-king-jena]

comment:1

The problem was in the comparison of poset elements. Comparing a poset element with any object that has no parent (for example, comparison with a Python int) resulted in an error.

I thought the straight forward way to fix it is: Use sage.structure.element.parent(bla) instead of bla.parent().

With the patch, plotting the Hasse diagram works, and the bug fix is doctested. Needs review, then!

[simon-king-jena]

Author: Simon King

[seblabbe]

comment:2

Great!

I tested it (on sage-4.7.2) and it solves the problem. All tests pass on sage/combinat repository. Documentation builds fine. Positive review.

But now, I get another problem. When the package sage -i dot2tex is installed the method layout_graphviz gets called by plot and it seems broken for another reason. But this should be another ticket :

sage: H.plot()
Traceback (most recent call last):
...
AttributeError: 'str' object has no attribute 'iteritems'

which is also obtained by doing

sage: H.layout_graphviz()    # require dot2tex
Traceback (most recent call last):
...
AttributeError: 'str' object has no attribute 'iteritems'

[seblabbe]

Reviewer: Sébastien Labbé

[seblabbe]

comment:3

But now, I get another problem.

This is now #12364.

[seblabbe]

comment:4

Same lines are edited by #11559. One will have to adapt to the other one.

[seblabbe]

comment:5

Changed the ticket title and its description since the original problem (hasse diagram ploting) might be (is?) solved indirectly by #10998.

[nthiery]

comment:6

Replying to @seblabbe:

Changed the ticket title and its description since the original problem (hasse diagram ploting) might be (is?) solved indirectly by #10998.

I removed the positive review, just in case we decide that this should be resolved as a duplicate of #10998.

[simon-king-jena]

comment:7

Replying to @nthiery:

Replying to @seblabbe:

Changed the ticket title and its description since the original problem (hasse diagram ploting) might be (is?) solved indirectly by #10998.

I removed the positive review, just in case we decide that this should be resolved as a duplicate of #10998.

Since #10998 does not touch the __eq__ method, I suggest that one first checks whether the bug triggered by comparing a poset element with a Python int is really fixed by #10998.

[simon-king-jena]

comment:8

Sorry, I can't test whether #10998 fixes the bug in comparison: It is in conflict with #11900, which is merged into sage-5.0.beta0

[simon-king-jena]

comment:9

I just tested that the error vanishes with #10998. From that point of view, one could suggest that the ticket be resolved as duplicate.

On the other hand, having parent(bla) instead of bla.parent() is another layer of security and won't hurt.

I leave the decision to you.

[nthiery]

comment:10

Replying to @simon-king-jena:

I just tested that the error vanishes with #10998. From that point of view, one could suggest that the ticket be resolved as duplicate.

On the other hand, having parent(bla) instead of bla.parent() is another layer of security and won't hurt.

I leave the decision to you.

I don't mind. In case we go for it, we might as well use "sage.structure.sage_object.have_same_parent".

[loefflerd]

Apply only this patch.

[loefflerd]

Changed reviewer from Sébastien Labbé to none

[loefflerd]

comment:11

Attachment: trac_12351-doctest.patch.gz

The new code from #10998 already uses have_same_parent. So all that's left to do here is to add a doctest to prevent the error happening again. I've uploaded a patch for that.

[loefflerd]

Changed author from Simon King to David Loeffler

[nthiery]

Reviewer: Nicolas M. Thiéry, Simon King

[jdemeyer]

Merged: sage-5.0.beta9