Compile Sage with Cython language_level=3str

[Jeroen Demeyer]

Cython 0.29 introduces a new language_level=3str, which is like language_level=3 except that strings are still str instead of unicode. It would be good to use this language_level in Sage to force Python 3 syntax in all Cython files.

[Erik Bray]

I am still confused by the whole language_level discussion. I tried following the discussion on the cython mailing list but I gave up.

If I understand correctly (and this is probably just restating what you wrote in the ticket description) language_level=3str means it's using Python 3 syntax in all Cython files, with the exception that unprefixed string literals are non-unicode strings on Python 2?

[Erik Bray]

Regarding your deleted comment from the other ticket (maybe it was meant for this one) about non-ASCII strings in string literals (there was an example you used with μ). If a docstring contains non-ASCII characters then I definitely think we should be explicitly using u''.

[Jeroen Demeyer]

Replying to embray:

I am still confused by the whole language_level discussion. I tried following the discussion on the cython mailing list but I gave up.

If I understand correctly (and this is probably just restating what you wrote in the ticket description) language_level=3str means it's using Python 3 syntax in all Cython files, with the exception that unprefixed string literals are non-unicode strings on Python 2?

It's not only about string literals. With language_level=3, the global str means unicode:

sage: cython('''
....: # cython: language_level=3
....: print(str)
....: ''')
<type 'unicode'>

[Jeroen Demeyer]

Replying to embray:

If a docstring contains non-ASCII characters then I definitely think we should be explicitly using u''.

On the other hand, I do consider it a bug that Cython disallows "μ" with language_level=3str since that's valid in both Python 2 and Python 3 (with a different meaning though).

[Erik Bray]

Replying to jdemeyer:

Replying to embray:

If a docstring contains non-ASCII characters then I definitely think we should be explicitly using u''.

On the other hand, I do consider it a bug that Cython disallows "μ" with language_level=3str since that's valid in both Python 2 and Python 3 (with a different meaning though).

Sure it's technically valid on Python 2, but not a great practice in most cases either, and with the different meaning there's a lot of potential for confusion around it so I think they're right to disallow it (even if it's unintentionally disallowed). But I also see your point.

[Erik Bray]

Replying to jdemeyer:

Replying to embray:

I am still confused by the whole language_level discussion. I tried following the discussion on the cython mailing list but I gave up.

If I understand correctly (and this is probably just restating what you wrote in the ticket description) language_level=3str means it's using Python 3 syntax in all Cython files, with the exception that unprefixed string literals are non-unicode strings on Python 2?

It's not only about string literals. With language_level=3, the global str means unicode:

sage: cython('''
....: # cython: language_level=3
....: print(str)
....: ''')
<type 'unicode'>

Ok, sure. But "3str" means str is just always str right? I guess I like that approach if so; it meshes well with the approach we've been already taking in Sage.

[Jeroen Demeyer]

Replying to embray:

Ok, sure. But "3str" means str is just always str right?

Yes, str is str and "literals" are also of type str.

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​bcba4f1	Upgrade to cypari2-1.3.1
​f77de1d	Upgrade to Cython 0.29
​0ea70b5	Compile Sage with Cython language_level=3str

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​25c25ee	Various fixes to rational_power_parts()
​5657422	Compile Sage with Cython language_level=3str
​b3a8a88	Remove __future__ statements from Cython files

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​732e8c5	Compile Sage with Cython language_level=3str
​172bb57	Remove __future__ statements from Cython files

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​7c36cdd	Explicitly set language_level=2 for a few modules
​8fb3196	Compile Sage with Cython language_level=3str
​b99e8da	Remove __future__ statements from Cython files
​b821f77	Doctest fixes

[Erik Bray]

Ticket retargeted after milestone closed (if you don't believe this ticket is appropriate for the Sage 8.8 release please retarget manually)

[Frédéric Chapoton]

red branch

[Jeroen Demeyer]

I know, but it's not ready for review anyway. This needs to wait for Cython 3.0

[Erik Bray]

As the Sage-8.8 release milestone is pending, we should delete the sage-8.8 milestone for tickets that are not actively being worked on or that still require significant work to move forward. If you feel that this ticket should be included in the next Sage release at the soonest please set its milestone to the next release milestone (sage-8.9).

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​dbcb78c	Explicitly set language_level=2 for a few modules
​0f789d8	Compile Sage with Cython language_level=3str
​aab046d	Remove __future__ statements from Cython files
​ae1f7f4	Doctest fixes

[Frédéric Chapoton]

red branch. I will review if you merge

[Erik Bray]

I think this just needs a rebase + testing again. This should be done as a dependency of #28000.

[Frédéric Chapoton]

branch merged with 8.9.rc0

---

New commits:

​5a33013

Merge branch 'u/jdemeyer/compile_sage_with_cython_language_level_3str' in 8.9.rc0

[Frédéric Chapoton]

Three patchbots seems to be happy, both with py2 and py3. Erik, would you approve a positive review ?

[Frédéric Chapoton]

Could please somebody else approve ? Or should I set to positive all by myself ?

[Frédéric Chapoton]

In the absence of any reaction whatsoever, I am setting to positive.

[Volker Braun]

merge conflict...

[Erik Bray]

Replying to chapoton:

Could please somebody else approve ? Or should I set to positive all by myself ?

Yeah why shouldn't you? You reviewed the change and tested it, didn't you? You don't need me to do that.

[git]

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

​0a34ee9

Merge branch 'public/ticket/26403' in 9.0.b1

[Frédéric Chapoton]

conflicts resolved, setting back to positive

[Volker Braun]

I'm getting the following failure on 32-bit:

File "src/sage/plot/plot3d/shapes.pyx", line 947, in sage.plot.plot3d.shapes.Sphere.get_grid
Failed example:
    Sphere(1).get_grid(100)
Expected:
    ([-10.0, ..., 0.0, ..., 10.0],
     [0.0, ..., 3.141592653589793, ..., 0.0])
Got:
    ([-10.0,
      -1.0471975511965976,
      -0.5235987755982988,
      6.1257422745431e-17,
      0.5235987755982989,
      1.0471975511965979,
      10.0],
     [0.0,
      1.0471975511965979,
      2.0943951023931957,
      3.141592653589793,
      4.188790204786391,
      5.235987755982989,
      0.0])
**********************************************************************
1 item had failures:
   1 of   3 in sage.plot.plot3d.shapes.Sphere.get_grid
    [162 tests, 1 failure, 43.51 s]
----------------------------------------------------------------------
sage -t --long src/sage/plot/plot3d/shapes.pyx  # 1 doctest failed
----------------------------------------------------------------------

[git]

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

​58a30a8

trac 26403 fix 32bit doctest

[Frédéric Chapoton]

ok, fixed