Support IPython for Shift+Enter (execute line in terminal)

[brettcannon]

Discussed in #16744

^{Originally posted by sergeyf July 20, 2021}

Environment data

• VS Code version: 1.58.2
• OS and version: Windows_NT x64 10.0.19043
• Python version (& distribution if applicable, e.g. Anaconda): 3.9,1
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): conda
• Value of the python.languageServer setting: pylance-2021.7.4

Expected behaviour

Pressing Shift + Enter copies code to the terminal and executes it

Actual behaviour

Pressing Shift + Enter copies code to the terminal but it doesn't execute

Steps to reproduce:

EDIT: #18177 could resolve this.

[brettcannon]

Thanks for the feature request! We are going to give the community 60 days from when this issue was created to provide 5 👍 upvotes on the opening comment to gauge general interest in this idea. If there's enough upvotes then we will consider this feature request in our future planning. If there's unfortunately not enough upvotes then we will close this issue.

[sergeyf]

Thank you!

[ryan-feeley]

This issue is funny, because sometimes I get the desired behavior already, and sometimes I don't, depending on my conda environment.

Environment A from CONDA-FORGE:
This environment gives me the normal behavior where code is copied and properly executed:
conda create --name py37A --channel conda-forge python=3.7 ipython ipykernel

So if I highlight the following and do shift+enter:

x = 5

y = x**2

def squareIt(x):
    return x**2

The IPython console looks like this:

In [1]: x = 5

In [2]: y = x**2

In [3]: def squareIt(x):
   ...:         return x**2
   ...: 

In [4]: 

Environment B from DEFAULTS (Anaconda):
This environment gives me the bug behavior where code is copied but not executed:
conda create --name py37B --channel defaults python=3.7 ipython ipykernel

So if I highlight the following and do shift+enter:

x = 5

y = x**2

def squareIt(x):
    return x**2

The IPython console looks like this:

In [1]: x = 5
   ...: y = x**2
   ...: def squareIt(x):
   ...:     return x**2
   ...: 
   ...: 

I'm on Windows 10, and the latest install of VSCode, but no insiders channel, or similar.

I don't know if the difference in behavior is from the conda-forge channel having slightly newer packages, or from some deeper differences. Updating the prompt_toolkit package in Environment B from the conda-forge channel (which updates several dependencies) is enough to allow that environment to have the "no bug expected behavior".

[ryan-feeley]

Since it might not be clear, here is how I currently get SHIFT+ENTER to send code to IPython. I first run a dummy line of code and do SHIFT+ENTER. This opens a terminal, and runs (sometimes...) the code in the normal python REPL. Then I type exit() to exit the python REPL. At the prompt I type ipython to start the IPython REPL. Now subsequent uses of SHIFT+ENTER send the code to IPython. With the code automatically run, or not, depending on my conda environment, as mentioned in my previous comment.

I think you can achieve the same behavior of running in the IPython terminal by adding this to your settings.json:

"python.terminal.launchArgs": [
     "-m",
    "IPython"
    "--no-autoindent",
],

(Just for reference, it appears that this issue has been around for awhile, and inconsistently reproduced. Perhaps the conda channel choice is the key to understanding the different observations. #169; ipython/ipython#13054). You'll also see various 3rd party VSCode extensions that people have created to try to work around this issue.

[ryan-feeley]

There have been various hacks proposed on the github issues and SO adding delays, and the like, to cause IPython to actually run the commands sent with SHIFT+ENTER. Perhaps the differences I observe based on package channel are somehow causing needed delays so occur automatically, by say using an unoptimized pyqt, event loop, or something, in the background to process the REPL on IPython side.

[brettcannon]

Thank you to everyone who upvoted this issue! Since the community showed interest in this feature request we will leave this issue open as something to consider implementing at some point in the future.

We do encourage people to continue 👍 this issue as it helps us prioritize our work based on what the community seems to want the most.