fix(windows,desktop): sidecar launch fixes and dev mode support#558
Open
slvnlrt wants to merge 6 commits intospacedriveapp:mainfrom
Open
fix(windows,desktop): sidecar launch fixes and dev mode support#558slvnlrt wants to merge 6 commits intospacedriveapp:mainfrom
slvnlrt wants to merge 6 commits intospacedriveapp:mainfrom
Conversation
Contributor
WalkthroughThis PR adds Windows platform support throughout the codebase. Changes include a Windows-specific build step for sidecar executable copying, stderr collection and display in process management, platform-specific implementations for filesystem permissions and usage reporting, and refactored daemon/IPC functionality with Windows stubs that force foreground operation on Windows. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes 🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Author
|
@CodeRabbit review |
Contributor
✅ Actions performedReview triggered.
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@desktop/src-tauri/build.rs`:
- Around line 76-84: The build currently ignores failures copying the sidecar
(std::fs::copy) so the build can succeed with no spacebot.exe; update the
needs_copy block to check the Result from std::fs::copy and if it returns Err
(or if after the copy dest.exists() is false) immediately abort the build (e.g.,
panic! with a clear message including src and dest) instead of silently ignoring
the error, referencing the variables needs_copy, src, dest, and the
std::fs::copy call; ensure this check happens before the final
println!("cargo:rerun-if-changed={}", src.display()) so the build fails fast
when the sidecar is missing or cannot be copied.
In `@interface/src/components/ConnectionScreen.tsx`:
- Around line 47-60: stderrLines is currently unbounded and the UI paragraph
collapses newlines; when appending to stderrLines inside the spawnBundledProcess
handlers (onError/onClose callbacks used in the spawnBundledProcess call),
truncate the buffer to a fixed max (e.g., keep only the last N lines by shifting
or slicing after push) so it never grows indefinitely and continue to use
slice(-20) for the snapshot; also change the renderer that currently uses the
<p> at the render site (the element referenced around lines 206-207) to preserve
line breaks (e.g., render inside a <pre> or a container with CSS white-space:
pre-wrap) so the last-20-lines snapshot remains readable; apply the same
bounded-buffer and rendering change to the other stderr usage sites mentioned
(the similar blocks at the other spawnBundledProcess usage around lines 82-85).
In `@src/daemon.rs`:
- Around line 297-300: The is_running function currently always returns None,
disabling duplicate-start protection; implement the TODO by reading the daemon
PID file (via DaemonPaths' pid path) and, on Windows, trying to open the process
with appropriate rights (e.g., OpenProcess with
PROCESS_QUERY_LIMITED_INFORMATION) and checking liveness with GetExitCodeProcess
(or equivalent) to determine if the PID is active; if the process is alive
return Some(pid), if the PID is stale remove the pid file and return None, and
ensure all errors are handled gracefully (log/ignore and treat as not running)
so callers of is_running (the start guard) correctly detect a running instance.
- Around line 177-179: The daemon's cleanup task currently swallows errors by
using "let _ = std::fs::remove_file(&cleanup_socket)" (and the similar pattern
later around the second remove_file), which hides failures; change both
occurrences to inspect the Result and log failures instead of discarding them
(e.g., match or if let Err(e) = std::fs::remove_file(&cleanup_socket) { log an
error/warn with the path and e }); do this inside the tokio::spawn closure that
uses cleanup_rx.wait_for so the cleanup path reports concrete errors (use your
project's logging facility or tracing) rather than discarding them.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 33ea31f2-936c-474b-b53c-02745f5a8f98
⛔ Files ignored due to path filters (7)
.cargo/config.tomlis excluded by!**/*.tomlCargo.lockis excluded by!**/*.lock,!**/*.lockCargo.tomlis excluded by!**/*.tomldesktop/src-tauri/Cargo.tomlis excluded by!**/*.tomldesktop/src-tauri/capabilities/default.jsonis excluded by!**/*.jsondesktop/src-tauri/gen/schemas/capabilities.jsonis excluded by!**/gen/**,!**/*.json,!**/gen/**desktop/src-tauri/icons/icon.icois excluded by!**/*.ico,!**/*.ico
📒 Files selected for processing (7)
desktop/src-tauri/build.rsinterface/src/components/ConnectionScreen.tsxinterface/src/platform.tssrc/api/ssh.rssrc/api/system.rssrc/daemon.rssrc/main.rs
Comment on lines
+177
to
+179
| tokio::spawn(async move { | ||
| let _ = cleanup_rx.wait_for(|shutdown| *shutdown).await; | ||
| let _ = std::fs::remove_file(&cleanup_socket); |
Contributor
There was a problem hiding this comment.
Don't swallow daemon cleanup failures.
These are best-effort paths, but silently discarding every remove_file error makes stale PID/socket problems much harder to diagnose on the next launch.
Proposed fix
tokio::spawn(async move {
let _ = cleanup_rx.wait_for(|shutdown| *shutdown).await;
- let _ = std::fs::remove_file(&cleanup_socket);
+ if let Err(error) = std::fs::remove_file(&cleanup_socket)
+ && error.kind() != std::io::ErrorKind::NotFound
+ {
+ tracing::warn!(
+ %error,
+ path = %cleanup_socket.display(),
+ "failed to remove IPC socket on shutdown"
+ );
+ }
});
@@
fn cleanup_stale_files(paths: &DaemonPaths) {
- let _ = std::fs::remove_file(&paths.pid_file);
- let _ = std::fs::remove_file(&paths.socket);
+ for path in [&paths.pid_file, &paths.socket] {
+ if let Err(error) = std::fs::remove_file(path)
+ && error.kind() != std::io::ErrorKind::NotFound
+ {
+ tracing::warn!(%error, path = %path.display(), "failed to remove stale daemon file");
+ }
+ }
}As per coding guidelines, "Don't silently discard errors. No let _ = on Results. Handle them, log them, or propagate them. The only exception is .ok() on channel sends where the receiver may be dropped."
Also applies to: 267-269
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/daemon.rs` around lines 177 - 179, The daemon's cleanup task currently
swallows errors by using "let _ = std::fs::remove_file(&cleanup_socket)" (and
the similar pattern later around the second remove_file), which hides failures;
change both occurrences to inspect the Result and log failures instead of
discarding them (e.g., match or if let Err(e) =
std::fs::remove_file(&cleanup_socket) { log an error/warn with the path and e
}); do this inside the tokio::spawn closure that uses cleanup_rx.wait_for so the
cleanup path reports concrete errors (use your project's logging facility or
tracing) rather than discarding them.
Author
There was a problem hiding this comment.
Valid point. However, cleanup_stale_files is pre-existing Unix-only code that this PR only reorganized into a platform module without changing its logic. Logging cleanup failures there would be a separate improvement. Kept as-is to limit this PR's scope to the Windows/sidecar fixes.
Contributor
There was a problem hiding this comment.
@slvnlrt, understood — keeping pre-existing logic untouched to limit PR scope is a reasonable call. Would you like me to open a follow-up issue to track adding error logging to both cleanup_stale_files and the tokio::spawn socket removal, so it doesn't get lost?
🧠 Learnings used
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-21T18:23:15.278Z
Learning: Applies to src/**/*.rs : Don't silently discard errors. No `let _ =` on Results. Handle them, log them, or propagate them. The only exception is `.ok()` on channel sends where the receiver may be dropped.
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-21T18:23:15.278Z
Learning: Applies to src/agent/**/*.rs : For changes in async/stateful paths (worker lifecycle, cancellation, retrigger, recall cache behavior), include explicit race/terminal-state reasoning in the PR summary and run targeted tests in addition to `just gate-pr`.
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-08T05:48:17.751Z
Learning: Applies to **/*.rs : Never silently discard errors; handle them, log them, or propagate them (no let _ = on Results except .ok() on channel sends)
Contributor
There was a problem hiding this comment.
🧩 Analysis chain
Opening the follow-up issue now!
🏁 Script executed:
#!/bin/bash
gh issue create \
--repo spacedriveapp/spacebot \
--assignee slvnlrt \
--title "refactor(daemon): log errors in cleanup_stale_files and shutdown socket removal" \
--body "## Summary
In \`src/daemon.rs\`, two cleanup paths currently silently discard \`remove_file\` errors, making stale PID/socket problems harder to diagnose on the next launch:
1. **\`unix_impl::cleanup_stale_files\`** – uses bare \`let _ = std::fs::remove_file(...)\` for both the PID file and the socket file.
2. **The \`tokio::spawn\` shutdown task in \`unix_impl::start_ipc_server\`** – uses \`let _ = std::fs::remove_file(&cleanup_socket)\` after the shutdown signal is received.
Per the project coding guidelines (\`AGENTS.md\`): *Don't silently discard errors. No \`let _ =\` on Results. Handle them, log them, or propagate them.*
## Suggested fix
\`\`\`diff
tokio::spawn(async move {
let _ = cleanup_rx.wait_for(|shutdown| *shutdown).await;
- let _ = std::fs::remove_file(&cleanup_socket);
+ if let Err(error) = std::fs::remove_file(&cleanup_socket)
+ && error.kind() != std::io::ErrorKind::NotFound
+ {
+ tracing::warn!(
+ %error,
+ path = %cleanup_socket.display(),
+ \"failed to remove IPC socket on shutdown\"
+ );
+ }
});
fn cleanup_stale_files(paths: &DaemonPaths) {
- let _ = std::fs::remove_file(&paths.pid_file);
- let _ = std::fs::remove_file(&paths.socket);
+ for path in [&paths.pid_file, &paths.socket] {
+ if let Err(error) = std::fs::remove_file(path)
+ && error.kind() != std::io::ErrorKind::NotFound
+ {
+ tracing::warn!(%error, path = %path.display(), \"failed to remove stale daemon file\");
+ }
+ }
}
\`\`\`
## Context
- The \`cleanup_stale_files\` logic was pre-existing Unix-only code; this PR (`#558`) only reorganised it into \`unix_impl\` without changing its logic. Improving it was intentionally deferred to keep that PR's scope focused on Windows/sidecar fixes.
- Raised by `@coderabbitai` in PR `#558`: https://github.com/spacedriveapp/spacebot/pull/558#discussion_r3068529186
- Requested by `@slvnlrt`
"
Length of output: 76
The assignee couldn't be set due to integration permissions — retrying without it.
🏁 Script executed:
#!/bin/bash
gh issue create \
--repo spacedriveapp/spacebot \
--title "refactor(daemon): log errors in cleanup_stale_files and shutdown socket removal" \
--body "## Summary
In \`src/daemon.rs\`, two cleanup paths currently silently discard \`remove_file\` errors, making stale PID/socket problems harder to diagnose on the next launch:
1. **\`unix_impl::cleanup_stale_files\`** – uses bare \`let _ = std::fs::remove_file(...)\` for both the PID file and the socket file.
2. **The \`tokio::spawn\` shutdown task in \`unix_impl::start_ipc_server\`** – uses \`let _ = std::fs::remove_file(&cleanup_socket)\` after the shutdown signal is received.
Per the project coding guidelines (\`AGENTS.md\`): *Don't silently discard errors. No \`let _ =\` on Results. Handle them, log them, or propagate them.*
## Suggested fix
\`\`\`diff
tokio::spawn(async move {
let _ = cleanup_rx.wait_for(|shutdown| *shutdown).await;
- let _ = std::fs::remove_file(&cleanup_socket);
+ if let Err(error) = std::fs::remove_file(&cleanup_socket)
+ && error.kind() != std::io::ErrorKind::NotFound
+ {
+ tracing::warn!(
+ %error,
+ path = %cleanup_socket.display(),
+ \"failed to remove IPC socket on shutdown\"
+ );
+ }
});
fn cleanup_stale_files(paths: &DaemonPaths) {
- let _ = std::fs::remove_file(&paths.pid_file);
- let _ = std::fs::remove_file(&paths.socket);
+ for path in [&paths.pid_file, &paths.socket] {
+ if let Err(error) = std::fs::remove_file(path)
+ && error.kind() != std::io::ErrorKind::NotFound
+ {
+ tracing::warn!(%error, path = %path.display(), \"failed to remove stale daemon file\");
+ }
+ }
}
\`\`\`
## Context
- The \`cleanup_stale_files\` logic was pre-existing Unix-only code; this PR (`#558`) only reorganised it into \`unix_impl\` without changing its logic. Improving it was intentionally deferred to keep that PR's scope focused on Windows/sidecar fixes.
- Raised by `@coderabbitai` in PR `#558`: https://github.com/spacedriveapp/spacebot/pull/558#discussion_r3068529186
- Requested by `@slvnlrt`
"
Length of output: 52
@slvnlrt, the follow-up issue has been created: #560 — refactor(daemon): log errors in cleanup_stale_files and shutdown socket removal.
I wasn't able to assign it automatically (integration permission limitation), so you may want to self-assign it on GitHub.
🧠 Learnings used
Learnt from: slvnlrt
Repo: spacedriveapp/spacebot PR: 558
File: src/daemon.rs:297-300
Timestamp: 2026-04-11T20:13:19.328Z
Learning: In `src/daemon.rs` (Rust, spacebot), `windows_impl::is_running` intentionally always returns `None` as a documented TODO stub. Windows duplicate-start protection is handled in practice by HTTP server port-bind contention rather than PID file checks. Implementing OpenProcess/GetExitCodeProcess liveness is a separate future task. Do not flag this as a bug in future reviews.
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-21T18:23:15.278Z
Learning: Applies to src/**/*.rs : Don't silently discard errors. No `let _ =` on Results. Handle them, log them, or propagate them. The only exception is `.ok()` on channel sends where the receiver may be dropped.
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-21T18:23:15.278Z
Learning: Applies to src/agent/**/*.rs : For changes in async/stateful paths (worker lifecycle, cancellation, retrigger, recall cache behavior), include explicit race/terminal-state reasoning in the PR summary and run targeted tests in addition to `just gate-pr`.
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-08T05:48:17.751Z
Learning: Applies to **/*.rs : Never silently discard errors; handle them, log them, or propagate them (no let _ = on Results except .ok() on channel sends)
Comment on lines
+297
to
+300
| /// TODO(windows): implement PID file + OpenProcess liveness check | ||
| pub fn is_running(_paths: &DaemonPaths) -> Option<u32> { | ||
| None | ||
| } |
Contributor
There was a problem hiding this comment.
Windows duplicate-start protection is effectively disabled here.
src/main.rs uses spacebot::daemon::is_running as the only pre-start guard, so always returning None allows a second Windows launch to proceed until it dies later on port/file contention instead of being rejected up front.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/daemon.rs` around lines 297 - 300, The is_running function currently
always returns None, disabling duplicate-start protection; implement the TODO by
reading the daemon PID file (via DaemonPaths' pid path) and, on Windows, trying
to open the process with appropriate rights (e.g., OpenProcess with
PROCESS_QUERY_LIMITED_INFORMATION) and checking liveness with GetExitCodeProcess
(or equivalent) to determine if the PID is active; if the process is alive
return Some(pid), if the PID is stale remove the pid file and return None, and
ensure all errors are handled gracefully (log/ignore and treat as not running)
so callers of is_running (the start guard) correctly detect a running instance.
Author
There was a problem hiding this comment.
This is intentionally a documented TODO stub. Implementing is_running via OpenProcess + GetExitCodeProcess is a separate feature that requires its own testing. This PR focuses on getting the sidecar launch working on Windows. The duplicate-start scenario is already handled in practice by port contention (the HTTP server will fail to bind if another instance is running).
Contributor
There was a problem hiding this comment.
@slvnlrt, understood — the stub is clearly documented with TODO(windows) comments and the port-contention fallback provides a reasonable safety net for now. I'll mark this as acknowledged.
✏️ Learnings added
Learnt from: slvnlrt
Repo: spacedriveapp/spacebot PR: 558
File: src/daemon.rs:297-300
Timestamp: 2026-04-11T20:13:19.328Z
Learning: In `src/daemon.rs` (Rust, spacebot), `windows_impl::is_running` intentionally always returns `None` as a documented TODO stub. Windows duplicate-start protection is handled in practice by HTTP server port-bind contention rather than PID file checks. Implementing OpenProcess/GetExitCodeProcess liveness is a separate future task. Do not flag this as a bug in future reviews.
Note: Learnings are effective only in the context of similar code segments. To apply general coding guidelines, please configure review instructions. You can manage existing learnings in the app.
🧠 Learnings used
Learnt from: ibhagwan
Repo: spacedriveapp/spacebot PR: 392
File: src/config/watcher.rs:585-611
Timestamp: 2026-03-21T15:48:22.938Z
Learning: In `src/config/watcher.rs` and `src/main.rs` (Rust, spacebot), named Signal adapter instances are intentionally started independently of the root `signal_config.enabled` flag. The root enabled flag only gates the default Signal adapter; named instances are governed solely by their own per-instance `enabled` flag. This is by design and should not be flagged as a bug in future reviews.
Learnt from: CR
Repo: spacedriveapp/spacebot PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-03-21T18:23:15.278Z
Learning: Applies to src/agent/**/*.rs : For changes in async/stateful paths (worker lifecycle, cancellation, retrigger, recall cache behavior), include explicit race/terminal-state reasoning in the PR summary and run targeted tests in addition to `just gate-pr`.
This was referenced Apr 11, 2026
- Make daemonize and libc unix-only deps, add windows-sys for Windows - Split daemon.rs into unix/windows modules with cfg guards - Add Windows GetDiskFreeSpaceExW impl for disk usage in system.rs - Guard Unix permissions in ssh.rs with cfg(unix) - Force foreground mode on Windows in main.rs - Set target-dir to SSD for faster builds
- Add icon.ico generated from Spacebot-iOS-Default asset (required for Windows RC) - Rename desktop binary from Spacebot to spacebot-desktop to avoid NTFS case-insensitive collision with the spacebot sidecar binary
The sidecar is launched via command.spawn() not command.execute(), so the scope must be under shell:allow-spawn for Tauri to authorize it.
In dev mode, Tauri resolves sidecars relative to the desktop executable. The sidecar lives in src-tauri/binaries/ but needs to be next to the exe in the target dir. Copy it in build.rs with timestamp comparison to only recopy when the source is newer.
Previously, sidecar crashes showed only the exit code with no details. Add onStderr handler to BundledProcessHandlers and display the last 20 stderr lines in the error message when the process exits unexpectedly.
![tembo[bot]](https://avatars.githubusercontent.com/in/1223475?s=60&v=4){kind=small}
| ) | ||
| }; | ||
| if ok == 0 { | ||
| return Err(anyhow::anyhow!("GetDiskFreeSpaceExW failed")); |
Contributor
There was a problem hiding this comment.
Minor: include the OS error (and path) here so failures are actionable on Windows.
Suggested change
| return Err(anyhow::anyhow!("GetDiskFreeSpaceExW failed")); | |
| if ok == 0 { | |
| return Err(anyhow::anyhow!( | |
| "GetDiskFreeSpaceExW failed for {}: {}", | |
| path.display(), | |
| std::io::Error::last_os_error(), | |
| )); | |
| } |
| } | ||
| setSidecarState("idle"); | ||
| }, | ||
| onStdout: (line) => { |
Contributor
There was a problem hiding this comment.
The sidecar logs are useful, but it’d be nice to avoid always logging in production builds (and trimming trailing newlines keeps the UI snapshot cleaner).
Suggested change
| onStdout: (line) => { | |
| onStdout: (line) => { | |
| const cleanLine = line.trimEnd(); | |
| if (import.meta.env.DEV) console.log("[sidecar stdout]", cleanLine); | |
| // Look for the "HTTP server listening" log line | |
| if (cleanLine.includes("HTTP server listening")) { | |
| sawReady = true; | |
| setSidecarState("running"); | |
| // Point the app at localhost | |
| setServerUrl("http://localhost:19898"); | |
| } | |
| }, | |
| onStderr: (line) => { | |
| const cleanLine = line.trimEnd(); | |
| if (import.meta.env.DEV) console.log("[sidecar stderr]", cleanLine); | |
| stderrLines.push(cleanLine); | |
| if (stderrLines.length > 20) { | |
| stderrLines.splice(0, stderrLines.length - 20); | |
| } | |
| }, |
| ] | ||
| } | ||
| }, | ||
| "shell:allow-stdin-write", |
Contributor
There was a problem hiding this comment.
Nice tightening of the spawn scope. Quick thought: if shell:allow-stdin-write / shell:allow-kill support scoped allow-lists (like spawn does), it might be worth scoping those to the same sidecar entry as well, so the permission surface stays minimal.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
6 commits fixing issues encountered while testing the Tauri desktop sidecar ("Start Local Server") on Windows. Three independent bugs prevented the sidecar from launching, two of which could also affect macOS.
Context and scope
These fixes are against the
spaceuibranch (PR #555). The issues were found while testing the Windows desktop build.Detailed Changes
1. Server Windows port (cross-platform)
Cargo.tomldaemonizeandlibcmoved tocfg(unix)deps, addedwindows-sysfor Windowssrc/daemon.rsunix_impl/windows_implmodules with#[cfg]guardssrc/api/system.rsGetDiskFreeSpaceExWimplementation for disk usagesrc/api/ssh.rsset_permissionsguarded with#[cfg(unix)]src/main.rs2. Desktop Windows build support
desktop/src-tauri/icons/icon.icodesktop/src-tauri/Cargo.tomlSpacebot->spacebot-desktop(see bug 4 below)3. Fix sidecar scope (could affect all platforms)
Problem:
capabilities/default.jsonhas the sidecar scope undershell:allow-execute, butplatform.tscallscommand.spawn(). Tauri rejects the spawn because the scope only authorizesexecute.Error:
Scoped command binaries/spacebot not foundFix: Moved sidecar scope from
shell:allow-executetoshell:allow-spawn. Also removed the bareshell:allow-executewhich was overly permissive.4. Fix sidecar binary name collision (Windows-specific)
Problem: On NTFS (case-insensitive), the desktop binary
Spacebot.exeand sidecarspacebot.execollide when sharing the sametarget-dir.tauri-plugin-shellresolves sidecars by joining the binary name withcurrent_exe().parent(), so it finds the desktop binary instead of the server.Result: The desktop app launches itself as the sidecar with
start --foregroundargs, re-initializes Tauri, tries to register global shortcuts already claimed by the first instance, and panics with `PluginInitialization("global-shortcut", "HotKey already registered")``.Fix: Renamed desktop binary to
spacebot-desktopto avoid the collision.5. Auto-copy sidecar in dev mode (Windows-specific,
#[cfg(windows)])Problem: In dev mode, Tauri resolves sidecars relative to
current_exe().parent()(the target dir), not fromsrc-tauri/binaries/. The sidecar binary is absent from the target dir.Error:
The specified file was not found. (os error 2)Fix: Added logic in
build.rsto copy the sidecar binary frombinaries/to the target dir, with timestamp comparison to only recopy when the source is newer. Guarded with#[cfg(windows)]. Production builds are unaffected (the bundler handles sidecar placement).6. Capture sidecar stderr on crash
Problem: When the sidecar crashed, the UI showed only "Process exited with code X" with no diagnostic information.
Fix: Added
onStderrhandler toBundledProcessHandlersinplatform.ts.ConnectionScreen.tsxnow collects stderr and displays the last 20 lines in the error message.Test plan
cargo build --release,spacebot start --foreground)bun x tauri dev)Note
Windows sidecar launch fixes: This PR unblocks desktop sidecar spawning on Windows by addressing six critical issues. The server now supports Windows with proper daemon handling and system APIs. The desktop binary was renamed to avoid NTFS case-insensitivity collisions with the sidecar. Tauri capability scopes were corrected to allow
spawn()instead of justexecute(). Dev mode now auto-copies the sidecar binary to the target directory. Sidecar stderr is captured and displayed on crashes for better diagnostics. The capability scope fix applies across all platforms.Written by Tembo for commit 37f4485. This will update automatically on new commits.