upgrade to ipython 7

[fchapoton]

Upgrade ipython and related packages:

• https://pypi.org/project/ipython/ 7.13.0 (current as of 2020-06-05: 7.15.0)
• https://pypi.org/project/ipykernel/ 5.2.1 (current as of 2020-06-05: 5.3.0)
• https://pypi.org/project/backcall/ 0.1.0 (current as of 2020-06-05)
• https://pypi.org/project/jedi/ 0.17.0 (current as of 2020-06-05)
• https://pypi.org/project/prompt-toolkit/ 3.0.5 (current as of 2020-06-05)

All of these package versions support python >= 3.6 (note #29033).

Tarballs: see checksums.ini [upstream_url]. (To configure Sage to download from the upstream URLs, use ./configure --enable-download-from-upstream-url)

CC: @embray @jdemeyer @slel @kiwifb @timokau @antonio-rojas

Component: packages: standard

Keywords: upgrade

Author: Jonathan Kliem, John Palmieri, Antonio Rojas

Branch: e21a7b0

Reviewer: John Palmieri, Matthias Koeppe

Issue created by migration from https://trac.sagemath.org/ticket/28197

[fchapoton]

Branch: public/ticket/28197

[fchapoton]

comment:1

This is basically working, but maybe without preparsing..

---

New commits:

0d4f0c7

adaptation to ipython 7 : first tentative

[fchapoton]

Commit: 0d4f0c7

[jhpalmieri]

comment:3

Should this depend on #28190?

[fchapoton]

Dependencies: #28190

[jplab]

comment:7

Current issues (8.9.beta3):

sage: installed_packages()['ipython']                                                                                                                                                                              
'7.6.1'
sage: 2^3                                                                                                                                                                                                          
1
sage: 2**3                                                                                                                                                                                                         
8
sage: n=4                                                                                                                                                                                                          
sage: type(n)                                                                                                                                                                                                      
<class 'int'>

The preparsing is not working since it has been commented out.

[jplab]

comment:8

Trying to launch ipython one get:

jplabbe@blackbomb:~/sage$ sage -ipython
Traceback (most recent call last):
...
pkg_resources.DistributionNotFound: The 'jedi>=0.10' distribution was not found and is required by ipython

So one has to install the jedi module as well.

[antonio-rojas]

comment:11

A patch that makes sage work with ipython 7, including the preparser, is available at [1]. I haven't bothered to make it backwards compatible with older ipython, and it probably needs some polishing, feel free to take it from there.

[1] https://aur.archlinux.org/cgit/aur.git/tree/sagemath-ipython7.patch?h=sagemath-python3-git

[embray]

comment:12

(For lack of a better component; perhaps we should add a "REPL" component...?)

[jhpalmieri]

comment:15

This is Python 3 only. Are there specific plans for when we drop Python 2 support from Sage? Version 9.1?

[antonio-rojas]

comment:16

ipython 7.10 breaks many more tests due to it no longer sorting dicts. Some tests output is actually random now. Updated patch at https://aur.archlinux.org/cgit/aur.git/tree/sagemath-ipython7.patch?h=sagemath-git

[embray]

comment:17

Ticket retargeted after milestone closed

[mwageringel]

comment:18

The new jedi completion engine seems to have some problems. For example with IPython 7.6.1, invoking tab completion like

sage: 1 + <TAB>

causes the interpreter to freeze for about 1.5 minutes (apparently, a largish cache is created at ~/.cache/jedi/). After that, several deprecated functions are imported into global scope causing deprecation warnings to be printed, such as:

Importing absolute_igusa_invariants_kohel from here is deprecated. If you need to use it, please import it directly from sage.schemes.hyperelliptic_curves.invariants
See https://trac.sagemath.org/28064 for details.
  return getattr(handle.access, attribute)(*args, **kwargs)

This also seems to slow down exiting Sage.

---

As for the display of dictionaries during doctests, I suggest to open a separate ticket to remove the sorting. This can be done independently from the IPython upgrade, once Python 2 support is dropped by Sage (in 9.1 I assume).

[fchapoton]

comment:19

a preparation step at #28948

[mkoeppe]

Reviewer: John Palmieri, Matthias Koeppe

[kliem]

comment:69

As for the tests, there is no obvious failure. So at least installing seems to work everywhere. I didn't test any generated docker images to see how it actually looks though.

[jhpalmieri]

comment:70

I'm carrying out my threat to set a positive review ;)

Dear all: feel free to add yourself as a reviewer or an author (or both), as you see fit.

[jhpalmieri]

comment:71

For a future ticket, we could also upgrade to 7.16.1. I just tried, and it builds and passes tests on my OS X box. I want to get this ticket in as it stands, though.

[kiwifb]

comment:72

Replying to @jhpalmieri:

For a future ticket, we could also upgrade to 7.16.1. I just tried, and it builds and passes tests on my OS X box. I want to get this ticket in as it stands, though.

For the record, I am at 7.16.1 in sage-on-gentoo as well. Merging as it it now is the most important bit.

[isuruf]

comment:73

Looks like merge failed

[sagetrac-git]

Changed commit from 751864b to e21a7b0

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

e21a7b0

Merge tag '9.2.beta5' into t/28197/public/ticket/28197-reb

[mkoeppe]

Changed dependencies from #28190 #29042 #29428 #29658 #29774 to none

[vbraun]

Changed branch from public/ticket/28197-reb to e21a7b0

[EmmanuelCharpentier]

Changed commit from e21a7b0 to none

[EmmanuelCharpentier]

comment:78

This update broke sage-shell-mode support for Sage session in emacs.

[mwageringel]

comment:79

Preparsing multi-line strings also seems to be broken by this upgrade. See #30417.

[mkoeppe]

Changed author from Jonathan Kliem, ​John Palmieri, Antonio Rojas to Jonathan Kliem, John Palmieri, Antonio Rojas

[egourgoulhon]

comment:81

See #30928 for a possible follow-up.