⚡ spark-rust-wasm · TPC-DS

all 99 queries · real Spark engine · 100% in-browser GitHub

loading engine…

Queries …

Dataset tier …

loading tier index…

Run everything 99 queries against the live engine

Query

Result

Pick a query and run it.

Dataset

loading fixtures…

The engine binary and the TPC-DS fixtures are fetched once from this origin, then every query runs inside the wasm sandbox — data never leaves the tab. Three dataset tiers (tiny ~7 MB · medium ~12 MB · rich ~20 MB) load on demand: switching tier re-streams only the Arrow tables — the ~13 MB engine stays resident. Fixtures are dsdgen'd at sf=1 then FK-pruned to a deterministic entity-subset so the fixed mega-dimensions (customer_demographics 1.92M, time_dim 86K, date_dim 73K) collapse to only the rows the sampled facts reference. Built from crates/spark-wasm/examples/tpcds. See README.