Is there an existing issue for this?
Category of feature request
Transpiler
Problem statement
The llm-transpile CLI command currently uses a single default Switch configuration. When users want to try multiple conversion patterns (e.g., different output formats or comment languages), they must modify the configuration file each time, which prevents parallel execution of different conversion patterns.
Switch supports configuration options such as:
target_type: notebook (default), file, sdpcomment_lang: English (default), Japanese, etc.max_fix_attempts: Number of retry attempts
Proposed Solution
Add an optional --switch-config-path parameter to the llm-transpile command that accepts a path to a Switch configuration YAML file. When omitted, the default configuration is used (current behavior).
Example usage:
databricks labs lakebridge llm-transpile \
--input-source /path/to/input \
--output-ws-folder /path/to/output \
--source-dialect pyspark \
--switch-config-path /Workspace/Users/user@example.com/switch_config.yaml \
--accept-terms true
Benefits:
- Users can run multiple conversion patterns in parallel with different config files
- No need to modify a shared config file between runs
- Enables easier E2E testing of different configuration combinations
Additional Context
No response
Is there an existing issue for this?
Category of feature request
Transpiler
Problem statement
The
llm-transpileCLI command currently uses a single default Switch configuration. When users want to try multiple conversion patterns (e.g., different output formats or comment languages), they must modify the configuration file each time, which prevents parallel execution of different conversion patterns.Switch supports configuration options such as:
target_type: notebook (default), file, sdpcomment_lang: English (default), Japanese, etc.max_fix_attempts: Number of retry attemptsProposed Solution
Add an optional
--switch-config-pathparameter to thellm-transpilecommand that accepts a path to a Switch configuration YAML file. When omitted, the default configuration is used (current behavior).Example usage:
databricks labs lakebridge llm-transpile \ --input-source /path/to/input \ --output-ws-folder /path/to/output \ --source-dialect pyspark \ --switch-config-path /Workspace/Users/user@example.com/switch_config.yaml \ --accept-terms trueBenefits:
Additional Context
No response