Changelog

v0.9 · 16 Oct 2026 — SQL workbench, .duckdb files, folders as one table

Query, Excel → Parquet

Query keeps one session: temporary tables, SET and ATTACH now carry over between runs.
Scripts with several statements run one by one, with a result for each and the failing statement underlined.
A catalog panel lists the tables, views and databases you create, and autocomplete knows them too.
Open .duckdb database files (read-only).
Drop a folder, or combine several files, to query them as one table — Hive-style folders become columns.
Local files are read in place and remote Parquet is read with range requests, so multi-gigabyte files open without loading them into memory.
Parquet export from Query writes the full result with its real column types, compressed with Snappy, ZSTD or GZIP.
Optional “keep files between visits” keeps copies on your own device. Off by default.
Excel → Parquet accepts .xls and no longer fails on large integers or columns that change type further down.

Viewer, Chart, converters, site-wide

Dark mode, following your system setting or the toggle in the header.
Load a file once and use it in every tool, or drop a file anywhere on the site.
New viewer grid: per-column stats with click-to-filter, resizable, reorderable, pinnable and hideable columns, and keyboard selection with copy.
Exact types in the viewer: microsecond and nanosecond timestamps, time zones, dates, large decimals, and nested list, struct and map columns.
Chart: drag to zoom, and dashboard panels filter each other.
⌘K command palette.
Simpler navigation, and a format switcher on every converter.

Chart, Viewer

Chart builds a dashboard from any file automatically; click a panel to fine-tune it in the builder.
Open public datasets straight from the chart page.
The viewer scrolls through every loaded row instead of paging, and exports include the whole filtered view.
Clearer messages when a remote file can't be loaded.

Converters, Query

Converters show the first 50 rows before you download, with copy to clipboard for text formats.
New TSV → Parquet converter.
Parquet → SQL speaks PostgreSQL, MySQL, SQLite or ANSI, with typed CREATE TABLE statements.
Query gets a sortable results grid, a schema panel, example queries, loading from a URL, a formatter, cancel, and “open in chart”.
A sample file on every converter.

Site-wide

Chart, Query

New Chart page: bar, line, area, scatter, histogram, pie, heatmap and box plot from any Parquet file, with SVG and PNG export and eight sample datasets.
Query: “try a sample” loads a sample file and runs a starter query.

Query

Query several files at once and join them.
Schema-aware autocomplete, ⌘/Ctrl+Enter to run, an automatic row limit, query history and result export.
A new logo.

Query

Query gets a proper code editor: DuckDB syntax highlighting, keyword help and completions.

New tools, Viewer

Compare two Parquet files: schema, row counts and cell-level differences.
Edit cells, add or delete rows, rename or drop columns, and save a new Parquet file.
Optimize: see how a file is encoded, re-encode it, and compare sizes.
Column profiling, and a map preview for GeoParquet files.
Install parquet.to as an app.

Query, converters, Viewer

Query Parquet files with SQL, in your browser.
Convert CSV, JSON and Excel to Parquet, with a choice of compression.
Inspect schema, row groups, column statistics and compression.
Open a file from a URL.
Open several files as tabs, sort columns, and share links that keep your view.
More exports: NDJSON, TSV, Markdown and HTML. Large CSV, NDJSON and TSV exports stream straight to disk.

Viewer, converters

Open Parquet files in your browser: browse, search and filter rows, and see the file's metadata. Nothing is uploaded.
Convert Parquet to CSV, JSON, Excel, SQL and XML.