- Removed the code that sets the default model in the configuration
- This change is to address the issue where the default model was being incorrectly set
- Simplified the configuration migration process by removing unnecessary steps
- Added getCurrentFileInfo handler to retrieve the absolute path of the current file
- Added getIDEServicePort handler to retrieve the IDE service port
- Updated fileHandler.ts and handlerRegister.ts to include the new handlers
- Created getCurrentFileInfo.ts in ide_services/endpoints to implement the getCurrentFileInfo handler
- Updated services.ts in ide_services to include the getCurrentFileInfo handler
- Fix devchat API base URL to prevent illegal API base error
- Set default API base URL to https://api.devchat.ai/v1
- Call fixDevChatApiBase function in activate method
- Removed DevChat.InstallCommandPython from package.json commands array
- Commented out DevChat.InstallCommandPython execution in statusBarView.ts
- No longer triggers Python command installation in statusBarItem creation
- Updated the language model in code completion to starcoder2:15b for improved performance.
- Overhauled prompt creation to include 'recent edit documents', 'task', 'call function define', 'similar blocks', and 'symbol defines' for better context recognition in logs.
- Ensured fallback to 'nvidia/starcoder2:15b' if no model is specified, enhancing reliability.
- Enhanced parsing logic to handle multiple messages in one go
- Added splitting of incoming data chunks before processing
- Implemented more robust error handling for JSON parsing
- Remove unnecessary installation check codes from devchat-commands
- Streamline setup process for new users
- Maintain application functionality and stability
- Refactor devchat API configuration for clarity and consistency
- Ensure code completion fallbacks to devchat when configured
- Streamline the process for activating devchat server code completion
- Added a preCompletionItem to hold the current autocomplete item
- Refactored how autocomplete items are returned to prevent flickering
- Improved logic to combine current line prefix and response code effectively
- Add support for DevChat API as a code completion source
- Implement dynamic model selection for code completions
- Introduce error handling and logging for DevChat API requests
- Implemented filtering based on comment symbols followed by filenames
- Added filtering for comments indicating omitted code sections
- Adjusted prefix gathering for inline completion to consider previous lines
