perfetto-sql

Query Android Perfetto traces by translating natural-language intents into valid SQL and executing them via trace_processor. This extracts slice and thread performance data, such as total time spent in a specific thread, returning results in CSV format.

SPAN_JOIN in Perfetto SQL is a mechanism to combine interval data from different tables based on overlapping time ranges. You need it for trace analysis when calculating overlaps between slices or aggregating metrics across concurrent thread executions.

Yes, you can use natural language to investigate CPU frequency changes. The system converts your intent into Perfetto SQL, queries the trace file via trace_processor, and returns the exact CPU frequency or startup-request metrics requested.

To handle incomplete slice durations in Perfetto trace data, your SQL queries must account for the specific edge case where dur = -1. Correct query construction ensures these incomplete intervals do not skew your performance aggregations.

Yes, Perfetto SQL requires strict standard library schema and module validation to ensure correctness. This enforces safe idempotent object creation and correct module usage, preventing errors during trace_processor execution.