See the query run.

Joins pulse along foreign keys. Filtered rows fade. Groups collapse into results. Live tables in your browser, nothing to install.

Open the tracerHow it works
QueryTrace at the JOIN stage: wires connect VENUE_ID between the VENUE and SCREENING tables and matched rows are lit.
OPEN THE TRACER

Wires connect the key columns. Matched rows light up together.

Six stages. Every row accounted for.

A query is not one step. QueryTrace splits it into the stages the database actually runs, and shows what each one keeps, drops, or builds.

  1. FROM

    Every table the query names lights up. Their rows are the starting set; everything else dims.

  2. JOIN

    Wires connect the key columns and matched rows light together. Outer joins keep the unmatched rows, dashed in gold.

  3. WHERE

    The columns the condition reads turn gold. Rows that fail drop out, so you can count exactly what survived.

  4. GROUP BY

    Rows take the color of their bucket, then each bucket collapses into a single row. HAVING filters the buckets.

  5. SELECT

    The kept columns turn mint and everything else falls away. Aliases and expressions are evaluated here.

  6. ORDER BY

    The result settles into its final order. Scrub back to any earlier stage and the rows are still there.

Click a result. See where it came from.

Pin any row in the result and every source row that produced it lights up in mint, across all the tables and at every stage. Hover to preview, click to keep it.

It works in reverse too: pin a row in a table and watch whether it survives the WHERE, which group it lands in, and which result it ends up inside.

The result panel with Beacon Theater pinned, and the VENUE and RESERVATION rows that produced it outlined in mint.

Nineteen lessons, from SELECT to subqueries.

Each lesson loads a small database, runs one query, and suggests a change to try. Pick one and the tracer opens with it already running.

Foundations

1. Choose result columns2. Filter individual rows with WHERE3. Combine conditions4. Calculate and sort5. Match patterns and ranges6. Keep unique results with DISTINCT7. Label a combined text field8. Summarize a set of rows9. Find missing values10. Aggregate within groups11. Filter completed groups with HAVING

Combining data

12. Match related rows with JOIN13. Match tables through WHERE14. Preserve unmatched rows15. Trace a complete query pipeline16. Chain multiple relationships17. Give one table two roles18. Stack compatible results19. Feed one query into another

Yours to run, anywhere.

Nothing leaves this tab.

SQLite compiled to WebAssembly runs inside the page. No account, no server, no tracking. The code is open source under the MIT license.

Bring your own schema.

Paste CREATE TABLE and INSERT statements, or a Workbench export, and trace against your own data.

The schema dialog listing five preloaded databases and a Build Your Own editor.
QueryTrace on a phone: the schema canvas on top with the playback dock, and the query sheet below.

Fits a phone.

The canvas fills the screen and the query and results live in a sheet beneath it.

Write it in your dialect.

MySQL, PostgreSQL, SQL Server and Oracle queries are translated to SQLite before they run, and the rewrite is shown under the editor so you learn both spellings.

  • MySQL / MariaDB
  • PostgreSQL
  • SQL Server
  • Oracle
  • SQLite

Open the tracer and run a query.

Open the tracer