.changeset/curly-tables-stare.md (1)

1-5: Tidy up the changeset phrasing and add a link.

Reads well and follows frontmatter, but consider:

• Link the fixed issue for traceability.
• Minor wording: “monorepo setting” → “monorepo setup”.

Example:

-Fixed a bug where the Biome Language Server didn't correctly compute the diagnostics of a monorepo setting, caused by an incorrect handling of the project status.
+Fixed a bug where the Biome Language Server didn't correctly compute diagnostics in a monorepo setup due to incorrect handling of the project status. See #7371.

.changeset/spicy-things-bathe.md (1)

1-5: LGTM, consistent with guidelines.

Frontmatter, tense, and link look right. If you want consistency with the other changeset, mirror the phrasing (“in a monorepo setup”).

crates/biome_service/src/projects.rs (1)

268-277: Avoid unnecessary clones in set_nested_settings.

You can move path and settings into the map since they’re owned params, saving two clones.

-            let mut nested_settings = data.nested_settings.clone();
-            nested_settings.insert(path.clone(), settings.clone());
+            let mut nested_settings = data.nested_settings.clone();
+            nested_settings.insert(path, settings);

crates/biome_lsp/src/session.rs (1)

468-472: Preserve tracing field or attach it to the log event.

The span declares diagnostic_count but it’s no longer recorded. Either record the field or attach it to the info! call.

-        info!("Diagnostics sent to the client {}", diagnostics.len());
+        tracing::Span::current().record("diagnostic_count", diagnostics.len());
+        info!(diagnostic_count = diagnostics.len(), "Diagnostics sent to the client");

crates/biome_lsp/src/server.rs (1)

357-360: Suffix match for config detection looks right; consider broadening .editorconfig watch scope

The switch to ends_with fixes nested config detection. One follow‑up: our registered watches only track root .editorconfig per folder; with this broader detection, we’ll never reach this branch for nested .editorconfig changes. Consider registering "**/.editorconfig" for workspace folders so nested editorconfigs also trigger a refresh.

-                            FileSystemWatcher {
-                                glob_pattern: GlobPattern::Relative(RelativePattern {
-                                    pattern: ".editorconfig".to_string(),
-                                    base_uri: OneOf::Left(folder.clone()),
-                                }),
-                                kind: Some(WatchKind::all()),
-                            },
+                            FileSystemWatcher {
+                                glob_pattern: GlobPattern::Relative(RelativePattern {
+                                    pattern: "**/.editorconfig".to_string(),
+                                    base_uri: OneOf::Left(folder.clone()),
+                                }),
+                                kind: Some(WatchKind::all()),
+                            },

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

💡 Knowledge Base configuration:

• MCP integration is disabled by default for public repositories
• Jira integration is disabled by default for public repositories
• Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.

Learnt from: CR
PR: biomejs/biome#0
File: crates/biome_diagnostics/CONTRIBUTING.md:0-0
Timestamp: 2025-08-11T11:46:05.836Z
Learning: Applies to crates/biome_diagnostics/**/*.rs : Types implementing Diagnostic must also implement Debug (e.g., use #[derive(Debug, Diagnostic)])

Learnt from: CR
PR: biomejs/biome#0
File: crates/biome_service/CONTRIBUTING.md:0-0
Timestamp: 2025-08-11T11:53:15.299Z
Learning: Applies to crates/biome_service/src/workspace/watcher.tests.rs : Place watcher tests for workspace methods in src/workspace/watcher.tests.rs

Learnt from: CR
PR: biomejs/biome#0
File: crates/biome_service/CONTRIBUTING.md:0-0
Timestamp: 2025-08-11T11:53:15.299Z
Learning: Applies to crates/biome_service/../biome_lsp/src/server.tests.rs : Keep end-to-end LSP tests in biome_lsp’s server.tests.rs

Learnt from: CR
PR: biomejs/biome#0
File: crates/biome_service/CONTRIBUTING.md:0-0
Timestamp: 2025-08-11T11:53:15.299Z
Learning: Applies to crates/biome_service/src/workspace_watcher.rs : Keep workspace state in sync with the filesystem using WorkspaceWatcher; only active in daemon mode

Learnt from: CR
PR: biomejs/biome#0
File: crates/biome_diagnostics/CONTRIBUTING.md:0-0
Timestamp: 2025-08-11T11:46:05.836Z
Learning: Applies to crates/biome_diagnostics/**/*.rs : Use helper advice types from biome_diagnostics::v2 (CodeFrameAdvice, CommandAdvice, DiffAdvice, LogAdvice) when suitable

Learnt from: CR
PR: biomejs/biome#0
File: crates/biome_diagnostics/CONTRIBUTING.md:0-0
Timestamp: 2025-08-11T11:46:05.836Z
Learning: Applies to crates/biome_diagnostics/**/*.rs : Specify category and severity using #[diagnostic(...)] on the type or derive them from fields

crates/biome_lsp/src/handlers/text_document.rs (1)

34-50: Don’t drop didOpen while configuration/project are unresolved — defer it

Early-returning here orphans the document (no tracking, didChange becomes a no-op, diagnostics never appear). This was flagged earlier and still applies.

Minimal changes to demote log levels and explicitly defer the open for replay after config load:

     let status = session.configuration_status();
 
     let project_key = if status.is_loaded() {
         match session.project_for_path(&path) {
             Some(project_key) => project_key,
             None => {
-                error!("Could not find project for {path}");
+                debug!("Deferring didOpen: project not resolved yet for {path}");
+                // enqueue for replay once project mapping stabilises
+                session.defer_did_open(url.clone(), version, content.clone(), language_hint);
                 return Ok(());
             }
         }
     } else {
         if status.is_plugin_error() {
             session.client.show_message(MessageType::WARNING, "The plugin loading has failed. Biome will report only parsing errors until the file is fixed or its usage is disabled.").await;
         }
-        error!("Configuration could not be loaded for {path}");
+        debug!("Deferring didOpen: configuration not loaded for {path}");
+        // enqueue for replay after configuration becomes loaded
+        session.defer_did_open(url.clone(), version, content.clone(), language_hint);
         return Ok(());
     };

Follow-up (recommended):

• Implement Session::defer_did_open + a flush triggered when configuration_status transitions to loaded.
• Alternatively, await a short watch/notify before deferring (to cover race at startup).

Happy to draft the queue + flush plumbing in Session if you like.

#!/bin/bash
# Evidence gathering: ensure no existing replay/queue already handles didOpen before config load.
set -euo pipefail

# Look for any queue/defer/replay of didOpen
rg -n -C2 -g 'crates/**' -P '(queue|defer|replay).*(didOpen|did_open)|didOpen.*(queue|defer|replay)'

# Who reacts to configuration becoming loaded?
rg -n -C2 -g 'crates/**' -P 'configuration_status\(\)|is_loaded\(\)|on_configuration_loaded'

# Check if anything reattaches documents after config load
rg -n -C2 -g 'crates/**' -P 'insert_document\(|open_file\('

crates/biome_lsp/src/handlers/text_document.rs (2)

12-12: Use warn! for transient, non-fatal conditions (and import it)

Deferrals aren’t errors. Bring in warn! so you can right-size log levels below.

-use tracing::{debug, error, field};
+use tracing::{debug, warn, error, field};

---

45-47: Avoid warning spam; tweak wording

Guard this message so it’s shown once per session, not per file. Small copy tweak reads better.

-            session.client.show_message(MessageType::WARNING, "The plugin loading has failed. Biome will report only parsing errors until the file is fixed or its usage is disabled.").await;
+            session.client.show_message(
+                MessageType::WARNING,
+                "Plugin loading failed. Biome will only report parsing errors until the plugin is fixed or disabled in your configuration."
+            ).await;

Consider tracking a shown_once flag on Session to de-duplicate.

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

💡 Knowledge Base configuration:

• MCP integration is disabled by default for public repositories
• Jira integration is disabled by default for public repositories
• Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.