Method run

benchmarks/src/clickbench.rs:129–244

(self)

impl RunOpt {
    pub async fn run(self) -> Result<()> {
        println!("Running benchmarks with the following options: {self:?}");

        let query_dir_metadata = fs::metadata(&self.queries_path).map_err(|e| {
            if e.kind() == ErrorKind::NotFound {
                exec_datafusion_err!(
                    "Query path '{}' does not exist.",
                    &self.queries_path.to_str().unwrap()
                )
            } else {
                DataFusionError::External(Box::new(e))
            }
        })?;

        if !query_dir_metadata.is_dir() {
            return Err(exec_datafusion_err!(
                "Query path '{}' is not a directory.",
                &self.queries_path.to_str().unwrap()
            ));
        }

        let query_range = match self.query {
            Some(query_id) => query_id..=query_id,
            None => 0..=usize::MAX,
        };

        // configure parquet options
        let mut config = self.common.config()?;

        if self.sorted_by.is_some() {
            println!("ℹ️  Data is registered with sort order");

            let has_prefer_sort = self
                .config_options
                .iter()
                .any(|opt| opt.contains("prefer_existing_sort=true"));

            if !has_prefer_sort {
                println!(
                    "ℹ️  Consider using -c datafusion.optimizer.prefer_existing_sort=true"
                );
                println!("ℹ️  to optimize queries while maintaining parallelism");
            }
        }

        // Apply user-provided configuration options
        for config_opt in &self.config_options {
            let parts: Vec<&str> = config_opt.splitn(2, '=').collect();
            if parts.len() != 2 {
                return Err(exec_datafusion_err!(
                    "Invalid config option format: '{}'. Expected 'key=value'",
                    config_opt
                ));
            }
            let key = parts[0];
            let value = parts[1];

            println!("Setting config: {key} = {value}");