Can Biome's LSP server enforce per-request timeouts without client‑side cancellation?

Current Capability

No, the Biome LSP server cannot currently enforce per-request timeouts independently of the client. It relies entirely on the $/cancelRequest notification sent by the LSP client to abort in-flight operations.

Technical Explanation

The lack of independent server-side timeouts is due to how Biome handles its core processing tasks. While the Biome CLI provides a --timeout flag, this is a process-level kill switch. In contrast, the LSP server is a persistent process; killing it to stop a single request would crash the entire language server instance, forcing the IDE to restart the connection.

The Cancellation Bottleneck

Even when a client sends a cancellation request, Biome's responsiveness is limited by its execution model. Formatting and linting are performed as single-pass operations within the Rust runtime. Because the formatter does not currently yield control back to the event loop during the processing of a single file, the server cannot check for the cancellation signal until the current file's transformation is complete.

Workarounds and Mitigation

Since there is no native configuration for per-request timeouts in the LSP server, consider these alternatives for CI or high-latency environments:

• CLI for CI Pipelines: Avoid using the LSP server in CI environments. Use the Biome CLI with the --timeout flag. This ensures that if a process hangs, the entire task fails cleanly rather than leaving the workspace in a partially formatted state.
• Client-Side Timeouts: Configure your IDE or LSP client to send the $/cancelRequest after a specific threshold. While this won't stop the server immediately (due to the single-pass nature mentioned above), it prevents the client from hanging indefinitely.

Verification

To verify the current behavior in your environment, you can attempt to format an exceptionally large file and manually trigger a cancellation via your IDE's command palette. You will likely observe that the server only acknowledges the cancellation after the file processing has finished.

Diagnostic Detail Needed: Are you experiencing these timeouts on specific file types (e.g., massive JSON files) or across all JavaScript/TypeScript assets? This distinction determines if the bottleneck is in the parser or the formatter.