Python3 : Test ticket for zip()

[wluebbe]

This ticket is used for doctesting the Python 3 compatibility of zip(). It is not a real ticket and should not be merged.

[chapoton]

one step in #21568

[chapoton]

as of sage 7.5.rc2

running all doctests with "from future_builtins import zip" does not give any error

I propose therefore to close this ticket

EDIT: this needs to be checked very carefully.

[saraedum]

Could you elaborate on what you meant by "This needs to be checked very carefully."

[chapoton]

Yes, I can.

What I mean is the following: one should find a way to run all doctests with the py3 behaviour of zip activated both in doctests and in code. I have not been able to do that so far, as it seems that adding an import line to src/sage/all.py is not enough.

Same thing for range in #16081

[chapoton]

Here is a test branch, displaying the issue : how to trigger future zip behaviour in globally in the code ?

---

New commits:

​737c668

a test branch for zip, but not working as desired

[saraedum]

Could you patch globals()['__builtins__'].zip in sage.all?

[jdemeyer]

You should really think about putting from six.moves import zip in sage.all. That will affect everybody using Sage. Doing something like that should be explicitly discussed on sage-devel (and not in a thread called "how to check zip, map, range, print for Python3").

[chapoton]

Replying to jdemeyer:

You should really think about putting from six.moves import zip in sage.all. That will affect everybody using Sage. Doing something like that should be explicitly discussed on sage-devel (and not in a thread called "how to check zip, map, range, print for Python3").

This is a *f#### test ticket*, with milestone wontfix. It has never been my intention to make this enter sage. Could you please make a more useful contribution ?

[jdemeyer]

Was it really necessary to be so unpolite?

It seems that the only indication that it was a test ticket is the milestone. Nothing in the title nor the description was making it clear that this was a test ticket. So that could easily be overlooked.

[jdemeyer]

Replying to chapoton:

Could you please make a more useful contribution ?

After being insulted like that? Why would I?

[chapoton]

Sorry for my rough reaction. All my apologies. Early morning's bad mood.

Could you please, given all that you know that I don't, tell if you think that the goal of this ticket is reachable ?

[git]

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

​a843004	Merge branch 'public/16082' in 7.6.b1
​8795f70	trac 16082 ! test branch for zip! do not merge !

[chapoton]

ok, this works and discovers a lot of issues: two notable ones being

    for op,i in reversed(zip(ef, indices)):
    TypeError: argument to reversed() must be a sequence

return Factorization(zip(pols, exps), leading_coeff)
      File "/home/u1/chapoton/sage/local/lib/python2.7/site-packages/sage/structure/factorization.py", line 294, in __init__
        raise TypeError("x must be a list")

[chapoton]

see #22271 for a partial cure

[chapoton]

see #22350 for the next pill in the cure

[chapoton]

then #22385

[chapoton]

then #22457

[chapoton]

then #22519

[saraedum]

I find it always confusing that this "needs review". Can we leave it as "needs work" until all the zip problems are fixed?

[chapoton]

They are now fixed, or should be. This really needs review now (but I admit it was set to needs review before reaching this state).

[saraedum]

So is the question here now whether the patchbot is happy?

[chapoton]

Yes, exactly. And sadly it seems not to be the case. At least this one failure seems related and needs investigation:

sage -t --long --warn-long 75.6 src/sage/graphs/generic_graph.py
**********************************************************************
File "src/sage/graphs/generic_graph.py", line 6558, in sage.graphs.generic_graph.GenericGraph.longest_path
Failed example:
    g.longest_path(algorithm="backtrack").edges()
Expected:
    [(0, 1, None), (1, 2, None), (2, 3, None), (3, 4, None), (4, 9, None), (5, 7, None), (5, 8, None), (6, 8, None), (6, 9, None)]
Got:
    []

[chapoton]

see #22603 for maybe the last step

[chapoton]

green bot, which means that all zip problems have been taken care of

[chapoton]

can we please close this as duplicate, now that everything is in order ?

[tscrim]

Do you want to keep this ticket around as a general test ticket against regressions or do you have another methodology for that?

[chapoton]

I have no other methodology. Maybe it is better to keep it open so that patchbots can be launched on it. The patchbots should now be very reluctant to test closed tickets indeed.

[chapoton]

so let us rather put that into needs_info status

[chapoton]

I now propose to close this one. If somebody ever reintroduces a bad zip, we will see that when running all doctests with python3.

[chapoton]

Erik, could you please close this one ?