Answer: CSV

The question asks which file format truncates floating-point values when unloading data from Snowflake. According to the community discussion and Snowflake documentation, when floating-point number columns are unloaded to CSV or JSON files, Snowflake truncates the values to approximately (15,9) precision. However, the values are not truncated when unloading to Parquet files. The most upvoted comment (9 upvotes) clearly states that CSV truncates floating-point values, and this aligns with Snowflake's behavior where CSV and JSON formats have precision limitations for floating-point numbers during unloading. Option B (CSV) is therefore correct. Option C (Avro) is not supported for unloading in Snowflake, as mentioned in the discussion comments. Option D (Parquet) explicitly does not truncate values. Option A (ORC) is not mentioned in the context of truncation in the discussion.

Author: LeetQuiz Editorial Team