Add a button to stop a terminal process from the Viewer pane #10625

juliasilge · 2025-11-18T04:27:19Z

Addresses #7747

@:apps
@:viewer

This PR extends the UI comm for previewing a URL to optionally keep track of the "source" of where that URL may be coming from, for example, the terminal that may be running that process. If we have such a process ID, then we show a new button the Viewer pane UI that can be used to stop the process. It's the same button that we use on the console to stop running code there.

I tried a few hacky things that might allow us to show such a button without truly tracking the source, but it did NOT result in good behavior. We can do PRs to the Shiny extension and the Quarto extension so they use this new API and can also show buttons that will stop their processes as well.

We also can extend this to process that come from the language runtimes, but I did not do that here. Do people want two buttons in such a case? The button that would be running in the console and the button in the viewer? A little weird maybe? I say let's wait and see what folks think.

Release Notes

New Features

Added new button to interrupt a running app (such as Streamlit or FastAPI) from the Viewer pane

Bug Fixes

N/A

QA Notes

If you have an app supported by the builtin positron-run-app extension, when you press the "play" button, you should see a new button pop up on the Viewer pane.

If you click the button, it should stop the app from running (similar to if you stopped it from the terminal)
Having another terminal focused or active should not interfere with the button behavior; the button should act like it is connected exactly to the process running the app

Be aware that we do not have clearing set up in the Viewer pane, i.e. #6827, #9323, #5344. Adding the button here might make that feel more urgent.

…process from the Viewer pane

github-actions · 2025-11-18T04:27:37Z

E2E Tests 🚀
This PR will run tests tagged with: @:critical @:apps @:viewer

^readme ^{valid tags}

juliasilge · 2025-11-21T23:29:57Z

positron/comms/ui-frontend-openrpc.json

+							"runtime",
+							"terminal"


Can we think of any other source we want to track, other than something running in the terminal and something running in the console?

juliasilge · 2025-11-21T23:30:28Z

src/vs/workbench/contrib/positronPreview/browser/components/actionBars.css

 }
+
+.preview-action-bar .action-bar-button-icon.interrupt::before {
+	color: var(--vscode-errorForeground);


Same as the button in the console

juliasilge · 2025-11-21T23:31:13Z

src/vs/workbench/contrib/positronPreview/browser/components/urlActionBars.tsx

+			// Send Ctrl+C (SIGINT) to the terminal
+			sourceTerminal.sendText('\x03', false);
+
+			// Poll after 3 seconds to check if the process actually stopped


If you use this polling to dismiss the button, it feels quite laggy unfortunately. I think we need to dismiss the button immediately but set up the polling to show an error, in case something goes wrong.

src/vs/workbench/contrib/positronPreview/browser/components/urlActionBars.tsx

juliasilge · 2025-11-21T23:32:49Z

src/vs/workbench/services/languageRuntime/common/languageRuntimeUiClient.ts

 export interface IRuntimeClientEvent {
 	name: UiFrontendEvent;
-	data: any;
+	data: unknown;


This is from the 106 merge, with a new (?) no-explicit-any eslint rule.

juliasilge · 2025-11-21T23:33:53Z

eslint.config.js

+			// --- Start Positron ---
+			'src/positron-dts/**',
+			// --- End Positron ---


The 106 merge added a new (I think) no-explict-any eslint rule, but I don't think we should apply it to our API file, similar to how the VS Code API file is treated.

juliasilge · 2025-11-21T23:51:08Z

I'm having a fair amount of trouble with the tests on this branch, but it looks unrelated to me.

@midleman can you take a look and also review the test I added in 4d50df8?

midleman · 2025-11-22T01:31:54Z

I'm having a fair amount of trouble with the tests on this branch, but it looks unrelated to me.

@midleman can you take a look and also review the test I added in 4d50df8?

The test failures seem to be related to very slow app start up... 🤔 I noticed some new useEffects hooks in the code. Code that possibly be related and/or running on startup? I'll take a closer look at your new test on Monday! Figuring out the app lag is higher priority. 🫠

Also, I'd update your branch with main. I think Jon fixed the shiny app test failure. Let's see how the run goes after that!

jmcphers

Nicely done, this will be very helpful! I was worried that the interrupt wouldn't work on Windows, but it actually works great. I do wonder if we should clear the Viewer pane after successfully interrupting as otherwise the app seems to get a little confused about what just happened.

juliasilge · 2025-11-22T23:47:15Z

I did merge from main. I still see the same failures, which are the same as on #10731:

e2e tests failing with:

Fixture "app" timeout of 60000ms exceeded during setup.

integration tests failing with:

Uncaught Error: spawn /home/runner/work/positron/positron/extensions/positron-python/python-env-tools/bin/pet ENOENT

So seems like those failures are not related to my changes here.

juliasilge · 2025-11-23T00:14:56Z

I do wonder if we should clear the Viewer pane after successfully interrupting as otherwise the app seems to get a little confused about what just happened.

Yeah, that worked out pretty well in 3c36ebb, adding the same thing that would happen if you clicked the button that says "Clear the current URL". I do want to highlight that there is a bit of messiness right now around how many previews there can be, but if you are using the preview pane (where this button is), every time a new URL or HTML preview is opened, it clears all previous previews first anyway. I think this is a good option for now!

midleman · 2025-11-24T16:34:40Z

test/e2e/tests/apps/python-apps.test.ts

Test looks great! One small suggestion, instead of creating a separate test, I would suggest adding the interrupt-button checks to the end of the existing streamlit app test. That way we get the coverage without the extra setup/teardown cost.

HA, I tried to do this and then remember why I had it in a separate test from before! 😆

The first test runs the Streamlit app in the viewer, then pops it to the editor to run there

My new test runs the Streamlit app in the viewer, then stops it with the new button

We do in fact need two separate tests here.

midleman · 2025-11-24T16:35:19Z

I did merge from main. I still see the same failures, which are the same as on #10731:

e2e tests failing with:

Fixture "app" timeout of 60000ms exceeded during setup.

integration tests failing with:

Uncaught Error: spawn /home/runner/work/positron/positron/extensions/positron-python/python-env-tools/bin/pet ENOENT

So seems like those failures are not related to my changes here.

Yep, you are right, these are issues in main at the moment. 😭

This reverts commit 93fa098.

juliasilge · 2025-11-24T19:48:52Z

All the tests (including my new one) in test/e2e/tests/apps/python-apps.test.ts are passing, along with all the @:apps and @:viewer tests ✅

The failures I see here are the same as on main:

e2e tests failing with:

Fixture "app" timeout of 60000ms exceeded during setup.

integration tests failing with:

Uncaught Error: spawn /home/runner/work/positron/positron/extensions/positron-python/python-env-tools/bin/pet ENOENT

I'm going to go ahead and merge so we can keep moving forward.

juliasilge added 2 commits November 17, 2025 21:24

Add a new "source" to track where Viewer pane URLs come from

9c468d1

Use new "source" tracking to add a button to stop a running terminal …

96aed52

…process from the Viewer pane

juliasilge added 12 commits November 21, 2025 10:55

Merged origin/main into feature/viewer-pane-stop-button

4c4fb85

Update Python extension test for new API

85284f5

Make the button dismiss immediately

2b8afab

Update positron-run-app to use new API

2db6bca

Revert change to positron-viewer (not very applicable)

cfc3f28

Changes to set up runtime source attribution

6c6e6d1

Remove this (not applicable after all)

4763cff

Whitespace

5b69409

Start work on updating enum

ca4fc62

Fix up positron.d.ts, for both enum and no-explicit-any eslint rule

d9df20d

Merged origin/main into feature/viewer-pane-stop-button

7a47093

Add draft of test for viewer interrupt button

4d50df8

juliasilge commented Nov 21, 2025

View reviewed changes

juliasilge marked this pull request as ready for review November 21, 2025 23:50

juliasilge requested review from jmcphers and midleman November 21, 2025 23:51

juliasilge changed the title ~~WIP: Add a button to stop a terminal process from the Viewer pane~~ Add a button to stop a terminal process from the Viewer pane Nov 21, 2025

jmcphers previously approved these changes Nov 22, 2025

View reviewed changes

Merged origin/main into feature/viewer-pane-stop-button

e8b6b5c

Clear the viewer pane once the process stops

3c36ebb

juliasilge dismissed jmcphers’s stale review via 3c36ebb November 23, 2025 00:11

midleman reviewed Nov 24, 2025

View reviewed changes

juliasilge added 3 commits November 24, 2025 09:43

Move new testing to existing streamlit test

93fa098

Merged origin/main into feature/viewer-pane-stop-button

28c4913

Revert "Move new testing to existing streamlit test"

91110c9

This reverts commit 93fa098.

juliasilge merged commit ee9f160 into main Nov 24, 2025
30 of 32 checks passed

juliasilge deleted the feature/viewer-pane-stop-button branch November 24, 2025 19:49

github-actions bot locked and limited conversation to collaborators Nov 24, 2025

Add a button to stop a terminal process from the Viewer pane #10625

Add a button to stop a terminal process from the Viewer pane #10625

Uh oh!

Conversation

juliasilge commented Nov 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Release Notes

New Features

Bug Fixes

QA Notes

Uh oh!

github-actions bot commented Nov 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

juliasilge Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

juliasilge Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

juliasilge Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

juliasilge Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

juliasilge Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

juliasilge commented Nov 21, 2025

Uh oh!

midleman commented Nov 22, 2025

Uh oh!

jmcphers left a comment

Choose a reason for hiding this comment

Uh oh!

juliasilge commented Nov 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

juliasilge commented Nov 23, 2025

Uh oh!

midleman Nov 24, 2025

Choose a reason for hiding this comment

Uh oh!

juliasilge Nov 24, 2025

Choose a reason for hiding this comment

Uh oh!

midleman commented Nov 24, 2025

Uh oh!

juliasilge commented Nov 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

juliasilge commented Nov 18, 2025 •

edited

Loading

github-actions bot commented Nov 18, 2025 •

edited

Loading

juliasilge commented Nov 22, 2025 •

edited

Loading

juliasilge commented Nov 24, 2025 •

edited

Loading