yaml-language-server Schema Configuration Not Working in Global Settings

[QiuminGe]

Issue Title: YAML Language Server Global Schema Configuration Not Working with Absolute Paths

Description

The YAML language server (yaml-language-server) fails to recognize custom JSON schemas when configured globally using absolute paths in the yaml.schemas setting within JupyterLab's LSP configuration. However, the same schema works correctly when specified per-file using the modeline comment (# yaml-language-server: $schema=...).

This indicates a path resolution issue where the global configuration does not properly handle absolute filesystem paths for schema associations, while relative paths or modeline directives function as expected .

Symptoms:
• No schema-based validation, autocompletion, or hover information occurs in YAML files when using absolute paths in global settings.

• The language server shows no error messages in logs, suggesting silent failure in schema loading.

• Switching to modeline comments immediately resolves the issue.

Reproduce

1. Configure Global Schema: In JupyterLab's LSP settings jupyterlab-settings

    {
        "language_servers": {
            "yaml-language-server": {
                "serverSettings": {
                    "yaml.schemas": {
                        "/home/jovyan/work/rules/demo/sigma-detection-rule-schema.json": "*.yaml"
                    },
                    "yaml.trace.server": "verbose"
                }
            }
        },
        "loggingConsole": "browser",
        "loggingLevel": "debug",
        "logAllCommunication": true,
        "setTrace": "verbose"
    }

2. Open a YAML File: Create or open a YAML file (e.g., test.yaml) that should match the pattern *.yaml.

3. Test Validation: Enter content that should trigger schema validation errors (e.g., invalid keys or types according to your schema). Observe that no validation errors or warnings appear.

4. Compare with Modeline: Add a modeline at the top of the same YAML file:

    # yaml-language-server: $schema=/home/jovyan/work/rules/demo/sigma-detection-rule-schema.json

Save the file. Now, schema validation errors should immediately appear, confirming the schema itself is valid and the server is functional .

Expected behavior

The global yaml.schemas configuration should correctly load and apply the JSON schema from the specified absolute path to all matching YAML files (*.yaml in this case), providing validation, autocompletion, and hover information seamlessly .

Specifically:
• Schema validation should trigger automatically for all YAML files matching the glob pattern.

• The behavior should be consistent with using the modeline comment, but applied globally to all relevant files.

• The language server should not silently ignore the configuration; any errors in path resolution should be logged.

Context

• Operating System and version: Linux (Containerized environment, e.g., JupyterPod)

• Browser and version: Not specified, but issue is backend/server-related.

• JupyterLab version: 4.4.9

• Language server and version: yaml-language-server v1.19.2 (installed globally via npm)

Required: installed server extensions

Required: installed lab extensions

Troubleshoot Output

Command Line Output

Browser Output (recommended for all interface issues)

Additional Context:
• The problem likely stems from how the yaml-language-server resolves paths in a workspace context. Documentation suggests that schema paths in global settings should often be relative to the project root rather than absolute . However, absolute paths should still be supported, especially when explicitly defined.

• This behavior creates a significant usability gap, as users must manually add modelines to every file instead of leveraging global configuration for project-wide schema management.

[QiuminGe]

• Advanced Settings Editor

• have no effect

• Configured separately it is ok

[QiuminGe]

• The same configuration works fine in VSCode.