Python 3 preparation: Change more integer divisions from / to // (part 2)

[wluebbe]

The attached patch is part of the enabling of Sage for Py2/3. Py2 integer division / is changed to P2/3 floor division // (see #15995). This ticket is similar to #16371 which already performed such changes.

see also #20471, #20483, #20468

[chapoton]

New commits:

​4c5457d

handling some more future divisions

[git]

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

​f9f15bb

some details

[git]

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

​3527969

wundo wrong change in crystals

[git]

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

​aaf1a12

another try for division in kirillov reshetikhin crystals

[jdemeyer]

The line

them does euclidean division (in Python2) or float division (in Python3)::

does not correspond to the code which does //.

[jdemeyer]

``2//3`` return ``0``.: return should be returns

[jdemeyer]

I feel that this ticket does too much at once. Most changes are non-trivial and require some work to review. It's too much for me anyway.

[chapoton]

should I split it in smaller chunks ? it is already only a small step in all the changes that have to be made...

[git]

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

​b5ab223

trac 18659 better corrections in rst files

[tscrim]

I don't think your changes in rigged_configurations/* are correct as they should all be Sage integers and, in certain cases, allowed to be rationals. Although perhaps it ends up being okay as there are no doctest failures, but I need to check (which I am doing now).

[tscrim]

Actually, it does seem to handle the division to rationals okay, but now it has values which are floats that should be integers.

[tscrim]

Okay, this should take care of the issues in rigged_configurations/*.

---

New commits:

​8de2be4

Fixing division in rigged configurations.

[jdemeyer]

Replying to chapoton:

should I split it in smaller chunks ? it is already only a small step in all the changes that have to be made...

I cannot speak for others, but for me it would improve the chances that I will review it.

[jdemeyer]

Splitting the ticket also means that we don't need an "all or nothing" approach. For example #15994 was rejected because of problems in one particular module. That is a pity, because the other changes are probably good.

[chapoton]

I have split into 2 tickets, the other one being #20533 for the combinat folder

---

New commits:

​11629f1	Merge branch 'u/tscrim/18659' into 7.2.b6
​ee97a51	splitting trac 18659 in smaller chunks

[chapoton]

I think only the changes in

*1 algebras/steenrod/steenrod_algebra_bases.py *2 graphs/generic_graph.py *3 modular/arithgroup/arithgroup_perm.py

really need to be checked seriously. All others are very simple.

For *1, there is a sentence that explain that all dimensions are divided by the given factor. Being dimensions, they surely must be integers.

For *2, the end result must be an integer (Wiener index), and we are simply dividing an even integer by 2.

For *3, it seems to be about cusp widths, that are in principle integers, as far as I understand.

[tscrim]

As far as I can tell, these are correct.

[chapoton]

Thanks a lot.

[chapoton]

This ticket has had seriously wrong consequences about the work done in #20533.

[chapoton]

This ticket's changes are plain wrong !

EDIT: can we easily undo this, quickly, please ??

[vbraun]

This is already in 7.3.beta0 so we can't undo it. You added 488db3dddfa94779c0ba612fd552239aec80ea5d that reverts some changes, so thats what you got.

[chapoton]

Yes, I guess that's my fault. Jeroen asked me to split a ticket, so I did it (as I could) by reverting some changes, and putting them in another ticket. Then the other ticket went in sage first, and this one after. And just undid everything done in the other ticket.

I will have to re-do the division clean-up in combinat. I am not happy about that.

[vbraun]

you can use git revert