upgrade electron to 32.2.5

[sharon-wang]

Summary

• addresses Crash when interacting with native OS toolbar on Linux #5927
• upgrades electron to v32.2.5, which contains the fix, based on chore: bump [email protected] microsoft/vscode#234160
• diff between v32.2.1 (current) and v32.2.5 (this pr): electron/electron@v32.2.1...v32.2.5

Release Notes

New Features

• N/A

Bug Fixes

• Fixes crash when clicking on 'File' in toolbar on Linux (Crash when interacting with native OS toolbar on Linux #5927)

Explanation

Running positron on Linux in a terminal via positron --verbose and then clicking the File option in the menubar results in Positron crashing. The terminal output shows the following error message:

[12657:0110/105954.066563:FATAL:check.cc(376)] Check failed: false. NOTREACHED log messages are omitted in official builds. Sorry!

This looks to be the same issue as microsoft/vscode#230604, which is ultimately due to a change in Chromium on Linux.

From microsoft/vscode#230604 (comment):

Update on this crash, it is triggered by the menu item File > New File... (command: welcome.showNewFileEntries) which has accelerator

vscode/src/vs/workbench/contrib/welcomeViews/common/newFile.contribution.ts

Line 34 in e62dc67

primary: KeyMod.Alt + KeyMod.CtrlCmd + KeyMod.WinCtrl + KeyCode.KeyN,
. The runtime on Linux cannot handle Win or Super keys for accelerator shortcut text and they always end up reaching this block https://source.chromium.org/chromium/chromium/src/+/main:ui/base/accelerators/accelerator.cc;l=413 which has not been a fatal call in chromium up until a recent runtime update (despite its name it only logs in release builds and in debug builds it will crash). Chromium has changed its implication in release builds starting with Electron 32 to crash as intended.
For long term, we should handle the shortcut text in Electron to avoid such crashes. Upstream issue electron/electron#44196

Electron v32.2.5 contains the fix for this issue.

QA Notes

Clicking on the File option in the native menubar on Linux should no longer crash Positron.

Since Electron has been updated by 4 patch versions, so it would be good to check that desktop builds across OS's are still working well!

This change fixes the issue for me in a local DEB release build on Ubuntu and also worked for a dev build on Linux for Wes 🎉

[github-actions]

E2E Tests 🚀  ^?
This PR will run tests tagged with: @critical

[sharon-wang]

Full test suite run: https://github.com/posit-dev/positron/actions/runs/12718317587

[petetronic]

Full test suite run: https://github.com/posit-dev/positron/actions/runs/12718317587

Could we ask QA about the win failure in the full run suite?

[sharon-wang]

@petetronic The test suite passed after running a third attempt.

Attempt 1: fail ❌
areas/new-project-wizard/new-project-r-jupyter.test.ts:38

Error: Timed out 30000ms waiting for expect(locator).toHaveCount(expected)

Locator: locator('text=/^Starting up|^Starting|^Discovering( \\w+)? interpreters|starting\\.$/i')
Expected: 0
Received: 1
Call log:
  - expect.toHaveCount with timeout 30000ms
  - waiting for locator('text=/^Starting up|^Starting|^Discovering( \\w+)? interpreters|starting\\.$/i')
    33 × locator resolved to 1 element
       - unexpected value "1"


   at ..\automation\src\positron\positronConsole.ts:137

  135 |
  136 | 		// ensure interpreter(s) containing starting/discovering do not exist in DOM
> 137 | 		await expect(page.locator('text=/^Starting up|^Starting|^Discovering( \\w+)? interpreters|starting\\.$/i')).toHaveCount(0, { timeout: 30000 });
      | 		                                                                                                            ^
  138 |
  139 | 		// ensure we are on Console tab
  140 | 		await page.getByRole('tab', { name: 'Console', exact: true }).locator('a').click();
    at PositronConsole.waitForReadyOrNoInterpreter (C:\a\positron\positron\test\automation\src\positron\positronConsole.ts:137:111)
    at C:\a\positron\positron\test\e2e\areas\new-project-wizard\new-project-r-jupyter.test.ts:14:38

Attempt 2: fail ❌

waitForReadyOrNoInterpreter seemed to fail each time

  Error: Timed out 30000ms waiting for expect(locator).toHaveCount(expected)

  Locator: locator('text=/^Starting up|^Starting|^Discovering( \\w+)? interpreters|starting\\.$/i')
  Expected: 0
  Received: 1
  Call log:
    - expect.toHaveCount with timeout 30000ms
    - waiting for locator('text=/^Starting up|^Starting|^Discovering( \\w+)? interpreters|starting\\.$/i')
      33 × locator resolved to 1 element
         - unexpected value "1"


     at ..\automation\src\positron\positronConsole.ts:137

    135 |
    136 | 		// ensure interpreter(s) containing starting/discovering do not exist in DOM
  > 137 | 		await expect(page.locator('text=/^Starting up|^Starting|^Discovering( \\w+)? interpreters|starting\\.$/i')).toHaveCount(0, { timeout: 30000 });
        | 		                                                                                                            ^
    138 |
    139 | 		// ensure we are on Console tab
    140 | 		await page.getByRole('tab', { name: 'Console', exact: true }).locator('a').click();

      at PositronConsole.waitForReadyOrNoInterpreter 

Attempt 3: pass ✅

[jmcphers]

FYI, the 1.96.0 merge (for which I have a PR open) updates Electron to 32.2.6, so maybe we don't need this if we can take that?

[petetronic]

That makes sense, @sharon-wang are you ok if we close these this time and rely on the v1.96 merge to update these instead?

[sharon-wang]

Sounds good! I'll close out this PR and #5953.

For linking purposes, the 1.96.0 merge PR: #5981