anomalyco/opencode
1
0
Fork 0
mirror of https://github.com/anomalyco/opencode.git synced 2026-07-22 02:05:26 -04:00
Code Issues 5.9k Packages Projects Releases 19 Wiki Activity

bug: ctrl+c exiting app even when prompt input isn't empty #3304

New Issue
Closed
opened 2026-02-16 17:39:35 -05:00 by yindo · 13 comments
yindo commented 2026-02-16 17:39:35 -05:00
Owner
Copy Link
Copy Source

Originally created by @rekram1-node on GitHub (Dec 4, 2025).

Originally assigned to: @kommander on GitHub.

Description

see screen recording:
https://discord.com/channels/1391832426048651334/1394667004979445931/1446206223396634715

OpenCode version

1.0.132

Operating System

Windows

Terminal

Windows Terminal

Originally created by @rekram1-node on GitHub (Dec 4, 2025). Originally assigned to: @kommander on GitHub. ### Description see screen recording: https://discord.com/channels/1391832426048651334/1394667004979445931/1446206223396634715 ### OpenCode version 1.0.132 ### Operating System Windows ### Terminal Windows Terminal
yindo added the windowshelp-wantedopentuibug labels 2026-02-16 17:39:35 -05:00
yindo closed this issue 2026-02-16 17:39:35 -05:00
yindo commented 2026-02-16 17:39:35 -05:00
Author
Owner
Copy Link
Copy Source

@github-actions[bot] commented on GitHub (Dec 4, 2025):

This issue might be a duplicate of existing issues. Please check:

  • #2999: Provide means to disable Ctrl-C (Windows users unable to use Ctrl+C without crashing)
  • #4997: Keybinds (explicitly lists: 'Windows ctrl+c quits app instead of copy')
  • #5070: exiting tui on windows has improper exit sequence handling (similar Windows terminal symptoms)
  • #3675: Keybind doesn't work in 1.0.7 (Windows 11, Windows Terminal - custom keybinds not working)
  • #4938: Keyboard shortcuts to edit prompts suck so much! (comprehensive keyboard/input issues)
  • #3213: Bug: opencode run and TUI exits hang on v0.15+ (Ctrl+C doesn't return shell prompt on Windows)

Feel free to ignore if none of these address your specific case.

@github-actions[bot] commented on GitHub (Dec 4, 2025): This issue might be a duplicate of existing issues. Please check: - #2999: Provide means to disable Ctrl-C (Windows users unable to use Ctrl+C without crashing) - #4997: Keybinds (explicitly lists: 'Windows ctrl+c quits app instead of copy') - #5070: exiting tui on windows has improper exit sequence handling (similar Windows terminal symptoms) - #3675: Keybind doesn't work in 1.0.7 (Windows 11, Windows Terminal - custom keybinds not working) - #4938: Keyboard shortcuts to edit prompts suck so much! (comprehensive keyboard/input issues) - #3213: Bug: `opencode run` and TUI exits hang on v0.15+ (Ctrl+C doesn't return shell prompt on Windows) Feel free to ignore if none of these address your specific case.
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Dec 7, 2025):

Hi! I’d like to work on this issue if it’s still available.
I'm able to reproduce the issue on Windows Terminal on both last release version and last commit version.

@val-2 commented on GitHub (Dec 7, 2025): Hi! I’d like to work on this issue if it’s still available. I'm able to reproduce the issue on Windows Terminal on both last release version and last commit version.
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Dec 7, 2025):

@val-2 that'd be great, note that opencode is built on opentui, so you may find that the bug lies at that lower level.

@rekram1-node commented on GitHub (Dec 7, 2025): @val-2 that'd be great, note that opencode is built on [opentui](https://github.com/sst/opentui), so you may find that the bug lies at that lower level.
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Dec 7, 2025):

Ok I think I found the issue but I'm not exactly sure how to resolve it.

Basically the bug starts manifesting when client.call("checkUpgrade",...) gets called, which happens 1 second after start.
I think the commands that get called for the check in some way "steal" ctrl-c handling of opencode and thus the normal exit path doesn't get followed but it just exits without clean-up. Might be partially related to #5070 as well.
So the reason why the bug happened in the video is not the start of the request but the start of checkUpgrade.
If I remove the checkUpgrade call the bug doesn't happen anymore.

Another thing I noticed is that the commands modify the name of the terminal tab when they get called, so it becomes "npm list", "npm config get registry"...

A possible fix I found is to use Bun.spawn instead of Bun shell $, for example:

$`npm list -g --depth=0`

becomes

Bun.spawn(["npm", "list", "-g", "--depth=0"])

With all the relative adjustments to get stdout and such.

I don't know enough about Bun internals to tell why this fixes the bug though. Or if it's more of a Bun issue than an OpenCode one.

Let me know if you'd like me to open a PR for the change or if you prefer a different approach.

@val-2 commented on GitHub (Dec 7, 2025): Ok I think I found the issue but I'm not exactly sure how to resolve it. Basically the bug starts manifesting when `client.call("checkUpgrade",...)` gets called, which happens 1 second after start. I think the commands that get called for the check in some way "steal" ctrl-c handling of opencode and thus the normal exit path doesn't get followed but it just exits without clean-up. Might be partially related to #5070 as well. So the reason why the bug happened in the video is not the start of the request but the start of checkUpgrade. If I remove the checkUpgrade call the bug doesn't happen anymore. Another thing I noticed is that the commands modify the name of the terminal tab when they get called, so it becomes "npm list", "npm config get registry"... A possible fix I found is to use Bun.spawn instead of Bun shell $, for example: ```typescript $`npm list -g --depth=0` ``` becomes ```typescript Bun.spawn(["npm", "list", "-g", "--depth=0"]) ``` With all the relative adjustments to get stdout and such. I don't know enough about Bun internals to tell why this fixes the bug though. Or if it's more of a Bun issue than an OpenCode one. Let me know if you'd like me to open a PR for the change or if you prefer a different approach.
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Dec 15, 2025):

Hi @rekram1-node, quick ping. Should I open a PR with this fix or try a different approach? Thanks!

@val-2 commented on GitHub (Dec 15, 2025): Hi @rekram1-node, quick ping. Should I open a PR with this fix or try a different approach? Thanks!
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Dec 15, 2025):

@val-2 so if you disable checkUpgrade the issue doesn't happen but if you enable it it does?

@rekram1-node commented on GitHub (Dec 15, 2025): @val-2 so if you disable checkUpgrade the issue doesn't happen but if you enable it it does?
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Dec 15, 2025):

Yes exactly, in general it seems to start happening when an "npm" command is launched with Bun.$
The bug stops happening when Bun.spawn or Bun.syncSpawn is used
Here is a video showing it

https://github.com/user-attachments/assets/efd1e2ed-74d3-482a-a5b1-df8f41f5c501

https://github.com/user-attachments/assets/fe7ca136-36bf-4a20-89b6-a205100b80c0

@val-2 commented on GitHub (Dec 15, 2025): Yes exactly, in general it seems to start happening when an "npm" command is launched with Bun.$ The bug stops happening when Bun.spawn or Bun.syncSpawn is used Here is a video showing it https://github.com/user-attachments/assets/efd1e2ed-74d3-482a-a5b1-df8f41f5c501 https://github.com/user-attachments/assets/fe7ca136-36bf-4a20-89b6-a205100b80c0
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Dec 15, 2025):

Also it seems to happen in particular with npm commands. If I replace $`npm config get registry`.text() with $`bun pm ls -g`.text() in the code in the second video, OpenCode exits correctly

@val-2 commented on GitHub (Dec 15, 2025): Also it seems to happen in particular with npm commands. If I replace ```$`npm config get registry`.text()``` with ```$`bun pm ls -g`.text()``` in the code in the second video, OpenCode exits correctly
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Dec 15, 2025):

Ohh interesting, yeah @val-2 go ahead w/ a PR thanks!

@rekram1-node commented on GitHub (Dec 15, 2025): Ohh interesting, yeah @val-2 go ahead w/ a PR thanks!
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@kommander commented on GitHub (Dec 20, 2025):

This is a bun issue that is reported and being worked on where it doesn't handle the windows CTRL_C_EVENT signal, so the signal cannot be intercepted and the process is immediately shut down.

@kommander commented on GitHub (Dec 20, 2025): This is a bun issue that is reported and being worked on where it doesn't handle the windows `CTRL_C_EVENT` signal, so the signal cannot be intercepted and the process is immediately shut down.
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@aiyolo commented on GitHub (Dec 26, 2025):

watching

@aiyolo commented on GitHub (Dec 26, 2025): watching
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Jan 4, 2026):

I tested on last commit and the bug on Windows still appears on the last Bun version. I had already tried experimenting with .quiet() and .nothrow() but it didn't work, also per Bun docs ".text() automatically calls .quiet()".

I also tried to find some other fixes, like re-enabling raw mode after npm commands finished running or using process.on("SIGINT", ...) to call the normal exit path, but I didn't find anything else that worked as a workaround while waiting for the fix in Bun, other than the method I sent earlier.

@val-2 commented on GitHub (Jan 4, 2026): I tested on last commit and the bug on Windows still appears on the last Bun version. I had already tried experimenting with `.quiet()` and `.nothrow()` but it didn't work, also per Bun docs ".text() automatically calls .quiet()". I also tried to find some other fixes, like re-enabling raw mode after npm commands finished running or using process.on("SIGINT", ...) to call the normal exit path, but I didn't find anything else that worked as a workaround while waiting for the fix in Bun, other than the method I sent earlier.
yindo commented 2026-02-16 17:39:36 -05:00
Author
Owner
Copy Link
Copy Source

@val-2 commented on GitHub (Jan 4, 2026):

Also since it may be useful for the bugfix upstream, by testing on a separate project it appears '\x03' handling with rawMode works correctly with Bun on Windows in normal conditions and breaks only after invoking npm/pnpm/yarn via Bun.$. The same issue occurs when using Bun.spawn with stdin: "inherit", but not with stdin: "ignore", so the bug might lie in inheritance and modification of the current TTY state.

@val-2 commented on GitHub (Jan 4, 2026): Also since it may be useful for the bugfix upstream, by testing on a separate project it appears '\x03' handling with rawMode works correctly with Bun on Windows in normal conditions and breaks only after invoking npm/pnpm/yarn via Bun.$. The same issue occurs when using Bun.spawn with stdin: "inherit", but not with stdin: "ignore", so the bug might lie in inheritance and modification of the current TTY state.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
acp
bad
bug
discussion
docs
documentation
enhancement
extension
good first issue
growth
help-wanted
mcp
model-problem
needs:compliance
nix
opentui
perf
pull-request
Mirrored from GitHub Pull Request
 tui
web
windows
zen
No Label bug help-wanted opentui windows
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#3304
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?