tursodatabase/turso · error · SystemExit

{run.path} has columns {run.columns}, {runs[0].path} has {co

Error message

{run.path} has columns {run.columns}, {runs[0].path} has {columns}

What it means

All CSV files given to the plotter are plotted as series on the same axes, so every run must share an identical column set. When a later run's columns differ from the first run's, the script exits naming both files and their column lists.

Source

Thrown at perf/tpc-h/plot/plot-tpch.py:69

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("csv_files", type=Path, nargs="+")
    parser.add_argument("--out", type=Path, default=Path("tpch"), metavar="PREFIX",
                        help="write PREFIX.png, PREFIX.pdf and PREFIX.tikz (default tpch)")
    parser.add_argument("--name", action="append", default=[], metavar="ENGINE=NAME",
                        help="legend name for an engine, e.g. limbo=Turso")
    args = parser.parse_args()
    names = dict(name.split("=", 1) for name in args.name)

    runs = [read_run(path) for path in args.csv_files]
    columns = runs[0].columns
    if not runs[0].rows or not columns:
        raise SystemExit("no results found")
    for run in runs[1:]:
        if run.columns != columns:
            raise SystemExit(f"{run.path} has columns {run.columns}, {runs[0].path} has {columns}")

    figure = Figure(runs, columns, names)
    for suffix in (".png", ".pdf"):
        output = args.out.with_suffix(suffix)
        figure.matplotlib(output)
        print(f"wrote {output}")
    output = args.out.with_suffix(".tikz")
    output.write_text(figure.tikz())
    print(f"wrote {output}")


class Run:
    """One CSV: the queries in file order and every engine's time for each."""

    def __init__(self, path, columns, rows):
        self.path = path
        self.columns = columns
        self.rows = rows

View on GitHub (pinned to 492c4a71cd)

Solutions

1. Regenerate all CSVs with the same benchmark version so schemas match.
2. Inspect the column mismatch: `head -1 old.csv` vs `head -1 new.csv`, and drop or convert the incompatible file.
3. Plot only runs from the same schema generation in one invocation.
4. If a column rename is expected, update the plotting script's expectations and migrate old CSVs.

Example fix

// before
python plot-tpch.py --csv-files old_schema.csv new_schema.csv -o out
// after
python plot-tpch.py --csv-files new_schema_run1.csv new_schema_run2.csv -o out

import csv
def columns_of(path):
    with open(path, newline="") as f:
        return next(csv.reader(f), [])
def schemas_match(paths):
    cols = columns_of(paths[0])
    return all(columns_of(p) == cols for p in paths[1:])

When it happens

Trigger: Running plot-tpch.py with multiple --csv-file/--name inputs produced by different benchmark versions, different scale factors, or a benchmark whose output columns changed between runs.

Common situations: Comparing results across git revisions after the benchmark's CSV schema changed; mixing old benchmark outputs with new ones; one CSV generated by a modified script with extra/renamed columns.

Understand the failure class

Background: Schema validation failed / invalid input schema: payload rejected because its shape doesn't match the expected schema — this error's family across 28 libraries.

Related errors

• no results found
• no query finished on any engine
• Unexpected row for INSERT
• pyturso's SQLAlchemy dialects require SQLAlchemy >= 2.0.45 (
• Cannot get raw connection from SQLAlchemy connection

AI-assisted analysis of tursodatabase/turso@492c4a71cd (2026-09-13). Data as JSON: /api/errors/567d79847c2c0aa7. Report an issue: GitHub.