Examples

Runnable scripts live in atlas-python/examples/. Each one is self-contained, writes to a temp directory, and runs in under a minute.

File	What it shows
`01_basics.py`	Create a store, define arrays, set attributes, reopen, read back. The minimal end-to-end loop.
`02_xarray.py`	Round-trip an `xr.Dataset` through atlas using both `atlas.add_xr_dataset(...)` and the `ds.atlas.write(...)` accessor.
`03_dask_streaming.py`	Stream a dask-chunked `xr.Dataset` into atlas one chunk at a time, preserving the chunk shape on disk; read back lazily.
`04_meta_formats.py`	Compare the six metadata format × compression combinations (`atlas.json`, `atlas.msgpack.zst`, …) on a 30-dataset store. Confirms auto-detection on reopen.
`05_codecs.py`	Compare `zstd` / `lz4` / `none` array codecs on a smooth `float32` field. Demonstrates per-array codec recording.
`06_stats_scan.py`	Find the dataset with the highest peak reading across a fleet of 32 sensors by scanning `array_stats` only — no raw data read.
`07_shared_arrays.py`	50 datasets sharing 2 physical array files. Prints the on-disk directory listing to confirm the layout.
`08_object_store.py`	Pass an obstore handle into `Atlas.create` / `Atlas.open` and run the full create / read / xarray / stats loop against it. Demos with `obstore.store.LocalStore` (no credentials); S3 / GCS / Azure variants are commented in at the top. Requires `pip install "atlas-python[cloud]"`.

Running

From a clone of the repository:

python atlas-python/examples/01_basics.py

Every script can be run standalone; there's no shared state between them. Each writes into a tempfile.TemporaryDirectory() that cleans up on exit.