Change all classic divisions to true divisions in combinat folder

[chapoton]

after #20468, let us take care of the rest of the combinat folder so that all test pass when running

export PYTHONIOENCODING="utf-8"
python -Qnew $(which sage-runtests) src/sage/combinat/*.py 

in a sage shell.

[chapoton]

New commits:

​4688e64

taking care of the remaining classic divisions in combinat folder

[jdemeyer]

General suggestion: add from __future__ import division on top of modules which have been fully converted. That way we know for sure that there are no old-style divisions left in that module.

[chapoton]

well, I understand, but I only took care of the failing doctests, not of the warnings (that often come from another part of sage, in fact)

[jdemeyer]

Replying to chapoton:

I only took care of the failing doctests, not of the warnings (that often come from another part of sage, in fact)

I really do not understand what you mean. In fact, it makes me doubt that you understood what I meant:

if you have removed all classic divisions from the combinat folder, I recommend to put from __future__ import division in the source code of the sage/combinat/*.py(x) files. Doing that will ensure that no old-style division can sneak in with future changes. It also makes it very explicit that no further work needs to be done with division in those modules.

[chapoton]

What I mean is the following.

With my branch, running python -Qnew $(which sage-runtests) src/sage/combinat/*.py gives no failing doctests.

Bur running python -Qwarnall $(which sage-runtests) src/sage/combinat/*.py still gives many warnings. Most of these warnings are related to classic division happening in other parts of sage, but there could be some still happening in /combinat without affecting the doctests.

I could try to add the from future import everywhere in all \combinat files and see if no problem occurs. Would this be a good idea ?

[jdemeyer]

Replying to chapoton:

Bur running python -Qwarnall $(which sage-runtests) src/sage/combinat/*.py still gives many warnings. Most of these warnings are related to classic division happening in other parts of sage

That does not matter.

but there could be some still happening in /combinat without affecting the doctests.

That could only happen if there are divisions which are not executed by any doctest. Let's hope this is not the case... in any case, it is hard to check otherwise.

I could try to add the from future import everywhere in all \combinat files and see if no problem occurs. Would this be a good idea ?

Yes, that is what I meant.

[chapoton]

Question: should I add the from future import in all of the combinat files, or only in the ones changed here or in #20468 ?

[jdemeyer]

I thought you checked all of them... in that case: all of them obviously.

[git]

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

​fb736f3

trac 20471 adding from future import division in some combinat files

[chapoton]

But many of them do not use division at all !

Wouldn't it be enough to just mark those where I did change something in the divisions ?

Or are we planning to have from future import division in every single python file in sage ?

[jdemeyer]

Replying to chapoton:

Or are we planning to have from future import division in every single python file in sage ?

This hasn't really been formally discussed. I would be in favour of having it in every module.

[chapoton]

I would prefer not to make a patchbomb right here.

We maybe can discuss on sage-devel the general question of adding from future import division everywhere.

Would you still agree to give a positive review for this ticket in its current state ?

[jdemeyer]

In alternating_sign_matrix, replace int(2) by 2.

[jdemeyer]

You changed this to really weird spacing: range(1,len(tst) // 2+1)

Usually, I don't mind spacing but I just noticed because you actually changed it.

[jdemeyer]

All fine except for the two comments above and the change in src/sage/combinat/perfect_matching.py (not because it is wrong, but because it is a non-obvious change that I still need to check).

[git]

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

​cbeeb2c

trac #20471 small details as suggested by reviewer

[tscrim]

The change in perfect_matching.py looks okay to me.

Is this ready for review again Frédéric?

[chapoton]

yes, please