microsoft / vscode-jupyter Goto Github PK

• Tests for "run current cell", "run all cells"

Open folder
Add python file
Add cells
Run cells
Close folder
Reopen
Run cells

Double cells end up being output

Are there plans to support a data explorer, or variables view where one would execute a command and have the ability to explore the variable or at least get some informations about the variable (type, min, max...etc). Similar to how Rstudio/Pycharm has.
if theres a manual solution like custom tree views via extension API i will be happy to try it.

No idea what I did, but I ended up with the following error: Cannot execute code, session has been disposed. What do i do next?

As a user, I'd expect to see a button that would restart the session or similar.

• Add support for bokeh
• Add support for integratepy
• Add tests

The progress message for Starting Jupyter must contain a spinner.
Basically we need to use VSC Api for progress messages to keep the UI for progress messages consistent.

Please use the API: IApplicationShell.withProgress

This would be better than a static message.
https://github.com/Microsoft/vscode-python/blob/master/src/client/interpreter/display/progressDisplay.ts#L45

Environment data

• VS Code version: Development
• Extension version (available under the Extensions sidebar): Development
• OS and version: Windows 10
• Python version (& distribution if applicable, e.g. Anaconda): Anaconda
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): conda
• Relevant/affected Python packages and their versions:

Actual behavior

If VSCode is not launched from an Anaconda prompt with the Scripts directory in the path then it will fail to start jupyter server.

Expected behavior

We should be able to start Jupyter server without starting code from an Anaconda command prompt.

Steps to reproduce:

1. Start VSCode from the taskbar or plain prompt
2. Attempt to run a code cell in jupyter
3. Jupyter not installed message pops up.

Logs

Output for Python in the Output panel (View→Output, change the drop-down the upper-right of the Output panel to Python)

Output from Console under the Developer Tools panel (toggle Developer Tools on under Help)

As sqlopsstudio displays the query result in spreadsheet format, It can be great to make table-like object (e.g, pandas.DataFrame, numpy.ndarray) rendered in similar way. AFAIK, Some IDEs already supported this feature, which makes inspecting data more efficient.

Actual behavior

Magic commands are not valid python commands, hence magic commands in the python file end up with red squigglys (marked as invalid syntax from Language Server).

Expected behavior

Add support for magic commands without noise in problems panel

Imagine somebody doesn't know about shift+enter auto advance
Instead they keep typing #%% and hitting run cell.

We should tell them about shift+enter or have a 'run and create new cell' entry

Currently we aren't passing the resource: Uri value into most methods.
This would result in various parts of the extension not working correctly in a multi-root environment (current code will pick settings from first available workspace folder).

Examples:

• https://github.com/Microsoft/vscode-python/blob/master/src/client/datascience/jupyterAvailability.ts#L22

This only applies to multi-root environments.

There is a web-based ipython console in the jupyterlab project, basically a QTConsole counterpart in browser.
It supports inline display of matplotlib plot and good pandas DataFrame rendering which is extremely important for interactive data processing related tasks.
hope to see that console embeded in VS Code.
a related issue may be (DonJayamanne/vscodeJupyter#19)

Feature requested via feedback:

if possible provide an interface to debug pandas dataframes.
Thank you very much

Python Interactive window should show a busy indicator

Preliminary spec:
/ icon spinning with "Starting Jupyter Server"

Repro steps:

1. Disconnect the machine from the internet.
2. "Run cell" on a markdown cell in a py file.

Expected:
Markdown cells are run and results are shown in the Python Interactive window.

Actual:
Result is empty. Code cells are run just fine though.

We're using prismjs to do colorization. This isn't as robust as textmate

Case in point,

On the left textmate is identifying import as keyword.control. On the right prism doesn't even have that in its grammar.

The solution would probably be:
Use https://www.npmjs.com/package/vscode-textmate to tokenize the source before we send it to react
Use our current solution to load the json files with the colors and turn them into a css
Expand this css to have colors for every token
Change the react code to turn the code into tokenized html.

Alternatively might use this:
https://www.npmjs.com/package/@uiw/react-codemirror

Then just give it the appropriate css files. That's what Jupyter uses.

• Persist history to temp storage with file hash
• Reopen temp storage on next vscode open
• Tests for temp storage

• Css/Output Cleanup
• Test cases for different cells/input
• Jupyter startup prior to pane
• Telemetry for history buttons

Environment data

VS Code version: 1.19.1
Python Extension version: 0.9.1
Python Version: 3.6.2 anaconda
OS and version: win10 version 1709 (build 17025.1000)

Actual behavior

ctrl+enter does nothing

Expected behavior

ctrl+enter executes "Run Selection/Line in Python Terminal" [ / python.execSelectionInTerminal ] context menu option and indexes cursor in editor to next line, i.e. like behavior with vscode R extension environment and with vs17 ide python script support.

Steps to reproduce:

• open .py file in vscode
• place cursor on a line of script
• enter "ctrl+enter"

Logs

Output from Python output panel

Output from Console window (Help->Developer Tools menu)

Actual behavior

When running a cell in python:

• Interactive window opens up
• Error message display about Jupyter not being installed

Expected behavior

• Do not display interactive window
• Display a better message with:
  • Descriptive error message
  • Ability to get more information (someone starting with VSC+Jupyter wouldn't know what to do - propbably a link to a page)

We'd probably want to capture telemetry on how many people tried this and didn't have Jupyter installed (so we can improve the getting started experience).

• Create a python file with the following code:

#%%
print("one")
#%%
print("two")

• Run all cells
• Save as notebook
• Open notebook in VSC and the resulting Python file has lots of extra whitespace

#%%
print("one")


#%%
print("two")

If I enter a cell that has an error and then I want to delete it, I have to scroll up past the error to do so. Perhaps the cell delete button could move with it?

We need something to show jupyter is starting up

Current spec:
/ Starting Jupyter Server with spinning '/' as shown in the status bar.

We should also start jupyter server as soon as we detect code cells.

Environment data

• VS Code version: 1.27.1
• Extension version: 2018.8.0
• OS and version: macOS 10.14
• Python version: Anaconda Python 3.6
• Type of virtual environment used: conda
• Relevant/affected Python packages and their versions: N/A

Actual behavior

When using a python file as a makeshift Jupyter notebook (with use of the VSCode Jupyter extension), valid ipython kernel code is flagged as invalid by the Python parser. For instance, cell magics, which begin with one or two percent signs, are flagged as syntax errors

Expected behavior

The parser should ignore valid ipython cell magic syntax; flagging these as problems just adds noise. A good solution might be to have the Python extension have its own internal Jupyter flag, which when enabled would understand that ipython cell magics are syntactically correct, but when disabled would treat the file as vanilla Python for the purpose of checking syntax.

Steps to reproduce:

Write e.g., %matplotlib inline at the top of the file.

• Add UI for collapsing output sections
• Unit and functional tests for collapsing input.

• Open a new editor window (new python file, but not saved to disk)
• Change language to Python (bottom right of VSC status bar)
• Run cell
• Click on the icon Go to code
• Does nothing

Actual behavior

Using keyboard to selecting the command Import Jupyter Notebook was displaying a Save Dialog.

Steps to reproduce:

• On the Interactive Window click the Save icon
• Using keyboard hit the esc key
• Using keyboard open the command palette and select Import Jupyter Notebook
• Hit the Enter key on keyboard
• You now get the Save Dialog

Cause - Save icon in Interactive Window still has focus.

Create a cell as follows:

#%% Wow
import sys
print(sys.executable)

I would expect the cell in the interative window to have a title of Wow.
Right now it displays the cell with the comments.

Currently we have the text #%% ... being displayed in the interactive window, only when there's text after #%%.

1. The height of the bar should match other VS Code tool windows (e.g. Output window)
2. The space b/w the icons should also match other VS Code windows
3. The icon colors need to be consistent (i'll check with Cherry)
4. Don't change the icon color on mouse hover.
5. Order of the icons (from left to right): Expand all, Collapse all, Restart kernel, SaveAs, Redo, Undo, Clear
6. Tooltip for each icon:
   -- Expand all: Expand all cells
   -- Collapse all: Hide input on all cells
   -- Restart kernel: Restart Jupyter kernel
   -- SaveAs: Save the results into a Jupyter Notebook file
   -- Redo: Redo the last operation
   -- Undo: Undo the last operation
   -- Clear: Clear all the content

• Run cell and wait for Interactive Widow to appear
• Reload VSC
• We now have an empty section where Interactive Window appeared

We should either not take up this space or display the Interactive Window with previous sessions results or other.

Open python file
add a cell
Bring up the context menu