With more details, I’d be happy to help you write a clear, accurate description or technical summary.
| Feature | Description | |---------|-------------| | | Up to 500 k rows/s from compressed CSV/Parquet on a 4‑core laptop. | | Dynamic Expressions | df.eval("log(colA) * sqrt(colB) + sin(colC)") compiled to native code on demand. | | Group‑by & Aggregations | Supports sum , mean , median , percentile , approx_quantile , plus user‑defined reducers. | | Streaming Mode | Process data that exceeds RAM by streaming blocks from disk with back‑pressure handling. | | GPU Offload (optional) | When a CUDA‑capable device is present, vectorised kernels are dispatched via Thrust. | | Built‑in Visualisation | Quick‑look plots ( df.plot(kind='hist') , df.scatter('x','y') ) using Matplotlib or Plotly back‑ends. | | Versioned Data | Immutable snapshots allow reproducible analyses across runs ( df.at_version('2024‑06‑01') ). | | Security | Sandboxed plug‑in execution (seccomp on Linux, Windows Job Objects) to prevent arbitrary code execution. | hsoda -030
gcc -shared -fPIC -o hello_plugin.so hello_plugin.c -I$(hsoda-config --include-dir) hsoda --load-plugin ./hello_plugin.so With more details, I’d be happy to help