fix(plan): rewrite `node_modules/.bin/*.cmd` to powershell on Windows by fengmk2 · Pull Request #345 · voidzero-dev/vite-task

fengmk2 · 2026-04-18T06:21:36Z

Summary

Running a node_modules/.bin/*.cmd shim on Windows triggers cmd.exe's "Terminate batch job (Y/N)?" prompt on Ctrl+C, which corrupts the terminal. At plan time, when which() resolves such a shim inside the workspace and a sibling .ps1 exists, rewrite the invocation to run the .ps1 directly through PowerShell — the spawn layer stays untouched and Ctrl+C propagates cleanly.

Closes voidzero-dev/vite-plus#1176

What the rewrite does

program_path: <abs path to pwsh.exe or powershell.exe>
args:         [-NoProfile, -NoLogo, -ExecutionPolicy, Bypass, -File,
               <.ps1 path, relative to task cwd>, ...original_args]

The .ps1 path is stored relative to the task's cwd (with / separators), so SpawnFingerprint.args is portable across machines — no absolute paths leak into cache keys. PowerShell resolves -File <relative> against its own working directory (the task's cwd) and lands on the correct file.

When it fires

Only when all of the following hold:

Extension is .cmd (case-insensitive) — cmd-shim never emits .bat.
Path lives inside the workspace root — global shims like %AppData%\npm\node_modules\.bin\foo.cmd are left alone.
Path's last two components are .bin / node_modules (case-insensitive).
A sibling .ps1 exists.
pwsh.exe or powershell.exe is on PATH.

Any miss and the original .cmd is kept — behavior matches pre-PR.

Tests

6 cross-platform unit tests in vite_task_plan::ps1_shim::tests cover the pure rewrite logic: workspace root, hoisted monorepo subpackage (.. traversal), missing sibling, non-.cmd extensions, .cmd outside .bin, .cmd outside the workspace.
Windows-only plan snapshot in plan_snapshots/fixtures/windows_cmd_shim_rewrite/ — a pnpm workspace with a sub-package. Two plan cases (dev_in_subpackage via cwd, dev_filter_from_root via --filter) produce byte-identical snapshots, proving the cache key doesn't depend on how the user navigates to the task. The harness gained a windows_only: bool flag on SnapshotsFile; redact_snapshot gained a pass that maps pwsh / powershell and their absolute host paths to <powershell> so the snapshot doesn't pin a specific runner image.

Test plan

cargo clippy --workspace --all-targets + cargo test --workspace green on macOS
Windows CI green
Manual on Windows: https://github.com/Curtion/report-vite-plus-1, Ctrl+C during vp run dev leaves terminal clean
Manual: confirm .cmd fallback on a Windows box with PATH stripped of PowerShell
Manual: confirm a globally installed .cmd outside the workspace is not rewritten

Note

Medium Risk
Changes Windows command resolution and cache fingerprint inputs for spawned tasks, which could affect execution behavior and caching; scope is gated to workspace-local .cmd shims with .ps1 siblings and is covered by targeted tests.

Overview
Avoids Windows Ctrl+C terminal corruption by rewriting workspace-local node_modules/.bin/*.cmd shims (with sibling .ps1) at plan time to run via pwsh.exe/powershell.exe -File <cwd-relative .ps1>; this changes the planned program_path/args and therefore the cache fingerprint to match what actually executes.

Adds ps1_shim rewrite logic with unit tests, integrates it into both normal spawn planning and synthetic task planning, and extends snapshot testing with a Windows-only fixture plus redaction/harness tweaks (including PowerShell host redaction and allowing fixture node_modules in git).

^{Reviewed by Cursor Bugbot for commit 24f2a8f. Configure here.}

Address review feedback on #345: - Extract the 5 fixed PowerShell prefix flags to a `const POWERSHELL_PREFIX` - Build `new_args` via iterator chain instead of manual Vec pushes - Drop the no-op `let _ = (&program_path, &args);` on non-Windows

Pulls in voidzero-dev/vite-task#345 which prefers .ps1 shims over .cmd on Windows to avoid the "Terminate batch job (Y/N)?" prompt and terminal corruption on Ctrl+C during `vp run dev`. Closes #1176

Picks up the cache-portability fix for voidzero-dev/vite-task#345 (PowerShell rewrite moved from plan layer to spawn layer).

Picks up voidzero-dev/vite-task#345 fix for missing `which` dep on Windows after the module move.

fengmk2 · 2026-04-18T08:03:01Z

@cursor review

Pulls in voidzero-dev/vite-task#345 which prefers .ps1 shims over .cmd on Windows to avoid the "Terminate batch job (Y/N)?" prompt and terminal corruption on Ctrl+C during `vp run dev`. Closes #1176

Picks up the cache-portability fix for voidzero-dev/vite-task#345 (PowerShell rewrite moved from plan layer to spawn layer).

Picks up voidzero-dev/vite-task#345 fix for missing `which` dep on Windows after the module move.

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 4f14ebe. Configure here.}

fengmk2 · 2026-04-18T13:15:12Z

before

after

Address review feedback on #345: - Extract the 5 fixed PowerShell prefix flags to a `const POWERSHELL_PREFIX` - Build `new_args` via iterator chain instead of manual Vec pushes - Drop the no-op `let _ = (&program_path, &args);` on non-Windows

fengmk2 · 2026-04-18T14:25:58Z

@cursor review

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 3e1fc38. Configure here.}

fengmk2 · 2026-04-18T14:48:23Z

@cursor review

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 26021ca. Configure here.}

fengmk2 · 2026-04-19T07:32:51Z

@cursor review

fengmk2 · 2026-04-19T07:50:04Z

@cursor review

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit be08530. Configure here.}

fengmk2 · 2026-04-20T02:45:35Z

@cursor review

fengmk2 · 2026-04-20T03:31:45Z

@cursor review

Address review feedback on #345: - Extract the 5 fixed PowerShell prefix flags to a `const POWERSHELL_PREFIX` - Build `new_args` via iterator chain instead of manual Vec pushes - Drop the no-op `let _ = (&program_path, &args);` on non-Windows

fengmk2 · 2026-04-20T06:27:56Z

@cursor review

Running a `node_modules/.bin/*.cmd` shim on Windows triggers `cmd.exe`'s "Terminate batch job (Y/N)?" prompt on Ctrl+C, which corrupts the terminal (backspace prints ^H, Ctrl+C prints ^C, input sluggish). At plan time, when `which()` resolves a `.cmd` shim inside the workspace and a sibling `.ps1` exists, rewrite the invocation to run the `.ps1` directly through PowerShell. The spawn layer stays untouched — it runs whatever the plan records, 1-for-1. Ctrl+C then propagates cleanly without the intermediate `cmd.exe` hop. The rewrite records: program_path: <abs path to pwsh.exe or powershell.exe> args: [-NoProfile, -NoLogo, -ExecutionPolicy, Bypass, -File, <.ps1 path relative to task cwd>, ...original_args] The `.ps1` path is stored relative to the task's cwd (with `/` separators), so `SpawnFingerprint.args` is portable across machines — no absolute paths leak into cache keys. PowerShell resolves `-File <relative>` against its own working directory (the task's cwd) and lands on the correct file. The rewrite fires only when all of the following hold: 1. Extension is `.cmd` (case-insensitive) — cmd-shim never emits `.bat`. 2. Path lives inside the workspace root — global shims like `%AppData%\npm\node_modules\.bin\foo.cmd` are left alone. 3. Path's last two components are `.bin` / `node_modules` (case-insensitive). 4. A sibling `.ps1` exists. 5. `pwsh.exe` or `powershell.exe` is on PATH. If any miss, the original `.cmd` is kept — behavior matches pre-PR. Tests ----- - 6 cross-platform unit tests in `vite_task_plan::ps1_shim::tests` cover the pure rewrite logic (root cwd, hoisted monorepo subpackage with `..` traversal, missing sibling, non-.cmd extensions, outside `.bin`, outside workspace). - Windows-only plan snapshot in `plan_snapshots/fixtures/windows_cmd_shim_rewrite/` — a pnpm workspace with a sub-package. Two plan cases (`dev_in_subpackage` via cwd, `dev_filter_from_root` via --filter) produce byte-identical snapshots, proving the cache key doesn't depend on how the user navigates to the task. The harness gained a `windows_only: bool` flag on `SnapshotsFile`; `redact_snapshot` gained a pass that maps `pwsh` / `powershell` and their absolute host paths to `<powershell>` so the snapshot doesn't pin a specific runner image. Closes voidzero-dev/vite-plus#1176

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 24f2a8f. Configure here.}

Pulls in voidzero-dev/vite-task#345 which prefers .ps1 shims over .cmd on Windows to avoid the "Terminate batch job (Y/N)?" prompt and terminal corruption on Ctrl+C during `vp run dev`. Closes #1176

Picks up the cache-portability fix for voidzero-dev/vite-task#345 (PowerShell rewrite moved from plan layer to spawn layer).

Picks up voidzero-dev/vite-task#345 fix for missing `which` dep on Windows after the module move.

fengmk2 self-assigned this Apr 18, 2026

fengmk2 mentioned this pull request Apr 18, 2026

chore: bump vite-task to e0229ffb voidzero-dev/vite-plus#1414

Draft

This comment was marked as resolved.

Sign in to view

cursor bot reviewed Apr 18, 2026

View reviewed changes

Comment thread crates/vite_task_plan/src/plan.rs Outdated

This comment was marked as resolved.

Sign in to view

fengmk2 added a commit to voidzero-dev/vite-plus that referenced this pull request Apr 18, 2026

chore: bump vite-task to c7523684

ceaac3f

Picks up the cache-portability fix for voidzero-dev/vite-task#345 (PowerShell rewrite moved from plan layer to spawn layer).

fengmk2 added a commit to voidzero-dev/vite-plus that referenced this pull request Apr 18, 2026

chore: bump vite-task to 7b74d616

8aa7edc

Picks up voidzero-dev/vite-task#345 fix for missing `which` dep on Windows after the module move.

This comment was marked as resolved.

Sign in to view

fengmk2 added a commit to voidzero-dev/vite-plus that referenced this pull request Apr 18, 2026

chore: bump vite-task to c7523684

bbe904f

Picks up the cache-portability fix for voidzero-dev/vite-task#345 (PowerShell rewrite moved from plan layer to spawn layer).

fengmk2 added a commit to voidzero-dev/vite-plus that referenced this pull request Apr 18, 2026

chore: bump vite-task to 7b74d616

249042f

Picks up voidzero-dev/vite-task#345 fix for missing `which` dep on Windows after the module move.

cursor bot reviewed Apr 18, 2026

View reviewed changes

fengmk2 changed the title ~~fix(plan): prefer .ps1 over .cmd shim on Windows~~ fix(spawn): prefer .ps1 over .cmd shim on Windows Apr 18, 2026

fengmk2 force-pushed the fix/prefer-ps1-over-cmd-shim branch from 5ce6c62 to 3e1fc38 Compare April 18, 2026 14:25

cursor bot reviewed Apr 18, 2026

View reviewed changes

fengmk2 changed the title ~~fix(spawn): prefer .ps1 over .cmd shim on Windows~~ fix: route node_modules/.bin .cmd shims through PowerShell on Windows Apr 18, 2026

cursor bot reviewed Apr 18, 2026

View reviewed changes

fengmk2 changed the title ~~fix: route node_modules/.bin .cmd shims through PowerShell on Windows~~ refactor: prefer .ps1 over .cmd shim on Windows Apr 18, 2026

fengmk2 changed the title ~~refactor: prefer .ps1 over .cmd shim on Windows~~ fix: prefer .ps1 over .cmd shim on Windows Apr 18, 2026

fengmk2 commented Apr 19, 2026

View reviewed changes

Comment thread crates/vite_task_plan/src/ps1_shim.rs Outdated

fengmk2 marked this pull request as ready for review April 19, 2026 07:32

fengmk2 requested a review from branchseer April 19, 2026 07:33

cursor bot reviewed Apr 19, 2026

View reviewed changes

Comment thread crates/vite_task_plan/src/ps1_shim.rs

cursor bot reviewed Apr 19, 2026

View reviewed changes

fengmk2 marked this pull request as draft April 20, 2026 02:17

cursor bot reviewed Apr 20, 2026

View reviewed changes

Comment thread crates/vite_task_plan/src/ps1_shim.rs

cursor bot reviewed Apr 20, 2026

View reviewed changes

Comment thread crates/vite_task_plan/tests/plan_snapshots/redact.rs Outdated

fengmk2 force-pushed the fix/prefer-ps1-over-cmd-shim branch 2 times, most recently from ea43983 to 24f2a8f Compare April 20, 2026 06:27

fengmk2 changed the title ~~fix: prefer .ps1 over .cmd shim on Windows~~ fix(plan): rewrite node_modules/.bin/*.cmd to powershell on Windows Apr 20, 2026

fengmk2 changed the title ~~fix(plan): rewrite node_modules/.bin/*.cmd to powershell on Windows~~ fix(plan): rewrite node_modules/.bin/*.cmd to powershell on Windows Apr 20, 2026

fengmk2 force-pushed the fix/prefer-ps1-over-cmd-shim branch from 24f2a8f to 892af62 Compare April 20, 2026 06:32

cursor bot reviewed Apr 20, 2026

View reviewed changes

fengmk2 marked this pull request as ready for review April 20, 2026 06:41

fengmk2 added a commit to voidzero-dev/vite-plus that referenced this pull request Apr 20, 2026

chore: bump vite-task to c7523684

5d9b792

Picks up the cache-portability fix for voidzero-dev/vite-task#345 (PowerShell rewrite moved from plan layer to spawn layer).

fengmk2 added a commit to voidzero-dev/vite-plus that referenced this pull request Apr 20, 2026

chore: bump vite-task to 7b74d616

2d09a07

Picks up voidzero-dev/vite-task#345 fix for missing `which` dep on Windows after the module move.

Conversation

fengmk2 commented Apr 18, 2026 • edited by cursor bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What the rewrite does

When it fires

Tests

Test plan

Uh oh!

This comment was marked as resolved.

Uh oh!

This comment was marked as resolved.

This comment was marked as resolved.

Uh oh!

fengmk2 commented Apr 18, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

fengmk2 commented Apr 18, 2026

Uh oh!

fengmk2 commented Apr 18, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

fengmk2 commented Apr 18, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

fengmk2 commented Apr 19, 2026

Uh oh!

Uh oh!

fengmk2 commented Apr 19, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

fengmk2 commented Apr 20, 2026

Uh oh!

Uh oh!

fengmk2 commented Apr 20, 2026

Uh oh!

Uh oh!

fengmk2 commented Apr 20, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

fengmk2 commented Apr 18, 2026 •

edited by cursor bot

Loading