Overview
This release reintroduces HikariCP for enhanced MySQL database performance and adds a public API for other plugins to interact with OllamaChat's AI chat functionality. It focuses on improving database efficiency, extensibility, and developer integration while maintaining compatibility with existing configurations.

New Features

• HikariCP Integration:
  • Added HikariCP connection pooling for MySQL databases to improve performance and resource management.
  • Configurable HikariCP settings in config.yml under database.mysql.hikari for tuning connection pool parameters (e.g., maximum-pool-size, connection-timeout).
• Public API:
  • Introduced OllamaChatAPI interface (com.ollamachat.api.OllamaChatAPI) for other plugins.
  • Supports sending AI queries, managing conversations, and accessing chat history programmatically with methods like sendAIQuery, createConversation, and listConversations.
  • Added API documentation in API.md for developer integration.

Enhancements

• Database Management:
  • Updated DatabaseManager to leverage HikariCP for MySQL connections, while preserving SQLite functionality.
• Dependency Management:
  • Updated DependencyLoader to include HikariCP (version 5.1.0) alongside MySQL and SQLite JDBC drivers, downloaded at runtime to plugins/OllamaChat/libs.
  • Maintained compileOnly in build.gradle for non-Paper dependencies to minimize JAR size.
• Configuration:
  • Added database.mysql.hikari section to config.yml with default settings for connection pooling.
  • Improved config.yml comments for clarity and ease of use.

Bug Fixes

• No specific bug fixes in this release, as the focus was on new features and performance improvements.

Notes for Upgrading

• Configuration:
  • Update config.yml to include the database.mysql.hikari section if using MySQL. Default values are added automatically.
  • Backup existing config.yml before updating to avoid data loss.
• Dependencies:
  • The plugin downloads HikariCP (version 5.1.0) automatically on first startup. Ensure server internet access or verify HikariCP-5.1.0.jar in plugins/OllamaChat/libs.
  • Delete the libs folder to force re-download if dependencies are corrupted.
• API Integration:
  • Developers can hook into OllamaChat using OllamaChatAPI. Add softdepend: [OllamaChat] to plugin.yml and refer to Github API.md.
• Compatibility:
  • This update is backward-compatible with existing configurations and databases.
• Testing:
  • Test database operations (e.g., /ollamachat conversation new) to verify MySQL/SQLite functionality.
  • Check server logs for dependency loading or database errors.

Known Issues

• HikariCP settings may require manual tuning for optimal performance based on server load and MySQL configuration.

Fixes
- Spigot and Bukkit Compatibility
New Features
- MySQL Database Support

Changes in this Update:

New Features

• Configurable Response Suggestion Count: Set the number of suggested responses via suggested-response-count (default: 3).
• Customizable Suggestion Prompt Template: Added suggested-response-prompt for tailored suggestion generation.
• Preset Suggested Responses: Introduced suggested-response-presets (e.g., "I see what you mean.") with toggle via suggested-response-presets-enabled.
• Model-Specific Suggestion Toggles: Control suggestion generation per model with suggested-response-model-toggles.
• Suggestion Hover Text: Added hover text for suggested responses to enhance interactivity.
• New Command: /ollamachat suggests toggle to enable/disable suggestions, with ollamachat.suggests.toggle permission.
• Suggestion Rate Limiting: Added suggested-response-cooldown (default: 10 seconds) to prevent suggestion spam.

Fixes

• Fixed Default Language Error: Corrected incorrect default language setting in the configuration file.

Configuration Updates

• New Configuration Options:
  suggested-response-count: 3
  suggested-response-prompt: "Conversation:\nUser: {prompt}\nAI: {response}\n\nBased on the above conversation, suggest {count} natural follow-up responses the user might want to say. They should be conversational in tone rather than questions. List them as:\n1. Response 1\n2. Response 2\n3. Response 3"
  suggested-response-presets:
  - "I see what you mean."
  - "That's interesting!"
  - "Tell me more about that."
  suggested-response-presets-enabled: false
  suggested-response-model-toggles:
  - llama3
  suggested-response-cooldown: 10
  
• Note: Backup existing configuration before updating to ensure compatibility with new structure.

Optimizations

• Improved suggestion generation logic for faster and more accurate responses.
• New permissions: ollamachat.suggests.toggle and ollamachat.suggests-presets.toggle.

Important Notes:

1. Configuration Changes:
   
   • New configuration options added for enhanced suggestion customization.
     
   • Recommended to back up old config and regenerate.
     
2. Delete lang directory to regenerate localization files.
3. Suggested responses and presets are disabled by default; enable via configuration.