python3 status

[Frédéric Chapoton]

Hello,

I have reached this point with the ugly python3 experimental branch "public/python3-experiment-8.1.b7":

┌────────────────────────────────────────────────────────────────────┐
│ SageMath version 8.1.beta7, Release Date: 2017-10-03               │
│ Type "notebook()" for the browser-based notebook interface.        │
│ Type "help()" for help.                                            │
└────────────────────────────────────────────────────────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Warning: this is a prerelease version, and it may be unstable.     ┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛
sage: 3*3
9
sage: parent(_)
Integer Ring

Cool, no ? Or maybe nobody cares ?

Many things are still not working. The cmp problem has been much reduced, but still not fully fixed. On our way is a large-scale unicode problem, and maybe another large scale hash problem.

Frederic

[John H Palmieri]

That's great! Thank you for all of your work on this.

  John

[Nils Bruin]

On Friday, October 13, 2017 at 5:56:11 PM UTC+2, Frédéric Chapoton wrote:

Hello,

I have reached this point with the ugly python3 experimental branch "public/python3-experiment-8.1.b7":

┌────────────────────────────────────────────────────────────────────┐
│ SageMath version 8.1.beta7, Release Date: 2017-10-03               │
│ Type "notebook()" for the browser-based notebook interface.        │
│ Type "help()" for help.                                            │
└────────────────────────────────────────────────────────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Warning: this is a prerelease version, and it may be unstable.     ┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛
sage: 3*3
9
sage: parent(_)
Integer Ring

Cool, no ? Or maybe nobody cares ?

It's definitely the former and not the latter. This is very cool and a lot of people care (or will, or should).
Once this branch gains traction, I imagine you may have to edit the "warning" banner and replace "may be" with "is" for a while.

[Andrew]

Definitely cool! Thanks for all of your work on this!
Andrew

[Jori Mäntysalo]

On Fri, 13 Oct 2017, Frédéric Chapoton wrote:

> Cool, no ? Or maybe nobody cares ?

Cool, yes!

--
Jori Mäntysalo

[Dima Pasechnik]

Given http://www.python3statement.org/ lists a number of crucial Sage dependencies to drop py2 on or before 2020, there is little choice. :-)

In fact, I think we should get Sage added to that list too, and get to work on py3 in the earnest.

[Julien Puydt]

Hi,

Le 13/10/2017 à 17:56, Frédéric Chapoton a écrit :
> Cool, no ? Or maybe nobody cares ?

Extremely cool!

> Many things are still not working. The cmp problem has been much
> reduced, but still not fully fixed. On our way is a large-scale unicode
> problem, and maybe another large scale hash problem.

Why are unicode and hash problems?

Snark on #sagemath

[John Cremona]

How can we be sure that new code witten by people (like me) who are
not python2/3 experts does not regress?

Also, I can see that projects which build on top of Sage will one day
have to make sure that their code is all python3-compliant (I am
thinking of the LMFDB). What's the simplest way of testing that now,
to see where the issues will lie?

John

On 13 October 2017 at 20:19, 'Julien Puydt' via sage-devel

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

[Frédéric Chapoton]

There is no simple and perfect way to really test that fully now.

Some things that can be done right now in the code of packages :

* check that print is used in a python3 compatible way

* check that cmp() is not used

* check that .iteritems, .iterkeys are not used

An useful diagnostic tool is "2to3" ; another one is "pep8" (to find deprecated syntax, for example syntax of raise).

Another way is to try to launch your module using "python -3".

The best way forward is to move as quickly as possible to a working python3+sage. We still have for example to make sage doctest system work with python3.

Frederic

https://stackoverflow.com/questions/40886456/how-to-detect-if-code-is-python-3-compatible

https://stackoverflow.com/questions/40815079/checking-code-for-compatibility-with-python-2-and-3

[Kwankyu Lee]

On Saturday, October 14, 2017 at 6:51:01 PM UTC+9, John Cremona wrote:

How can we be sure that new code witten by people (like me) who are
not python2/3 experts does not regress?

This would help. While coding on Sage, I refer to Python 3 reference manual :-)

[Volker Braun]

There are a few guards in place

* Doctests don't accept print statements any more

* There is a unit test that checks that code is syntactically Py3 

Thats not enough to catch .iteritems, for example, but we'll get better test coverage over time.

[Jeroen Demeyer]

On 2017-10-14 11:50, John Cremona wrote:
> How can we be sure that new code witten by people (like me) who are
> not python2/3 experts does not regress?

I think it's also important to mention that Cython is quite different
from plain Python in this regard. Cython generally tries to be
compatible with both Python 2 and Python 3. This means that some Python
2 only code like xrange() or dict.iteritems() works in Cython with
Python 3. Also, some Python 3 only code works on Python 2. For example,
keyword-only arguments, f-strings, new super(), yield from.

[Erik Bray]

Very cool. I still hope to help out more with this, especially the
string problems which I'm practiced at dealing with. I was starting
to several weeks ago when I tried to reproduce the Python 3 issues
from scratch, but then got side-tracked, and then went on a long trip,
so now I'll need to start over again on a Python 3 build.