Apache Arrow / Feather (.arrow / .feather)
.arrow.featherApache Arrow, Polars, PyArrow, pandas, DuckDB, and modern high-performance analytics engines
What it is
The standard open-source columnar in-memory and on-disk binary format for high-speed scientific and analytics data interchange. Read directly via the official apache-arrow library.
Who produces it
Data scientists and engineers using Python (Polars, pandas, PyArrow), R (arrow), Julia, Rust, or DuckDB to store columnar datasets without serializing to slow text formats.
What we read / what we don't read
What we read
- Arrow IPC File format (.arrow, .feather with ARROW1 magic header)
- Arrow IPC Stream format
- Numeric columns (Float32/64, Int8..64, Uint8..64) normalized to Float64Array
- Utf8 and LargeUtf8 string columns
- Boolean columns normalized to numeric binary flags
- Date32/Date64 and Timestamp columns normalized to epoch milliseconds (with microsecond and nanosecond unit scaling)
What we don't read
- Nested List, Struct, Map, and Union columns (skipped to preserve 2D spreadsheet topology)
- 64-bit integers exceeding 2^53 - 1 (may lose least-significant precision due to JavaScript Number limits)
How to import it into AltaiPlot
- Open AltaiPlot and drag your .arrow or .feather file onto the window, or use File → Import.
- AltaiPlot detects the Arrow IPC signature and parses columns with zero memory duplication.
- Preview the schema and column names.
- Click Import to load the columnar data instantly.
Auditability note: this page's capability claims are sourced fromsrc/renderer/src/engine/v7/import/importers/ArrowImporter.ts:45 — see the full format list for every extension AltaiPlot recognizes.