Configuration
Contents
Configuration¶
Core Dask has detailed documentation describing how configuration works in Dask. This page exists to cover configuration options specific to dask-awkward.
The file which defines the default configuration parameters is in the
dask-awkward repository at the path src/dask_awkward/awkward.yaml
.
The file attempts to be self documenting.
Top level table¶
These top level parameters are configuration under the awkward
namespace in Dask configuration format. For example, they can be set
with the form:
with dask.config.set({"awkward.<option>": value}):
...
raise-failed-meta
(default:False
): If this option is set toTrue
, then an exception will be raised if dask-awkward fails to automatically determine the metadata of a new collection as task graphs are built.compute-unknown-meta
(default:True
): In the event that dask-awkward cannot determine the metadata for a collection, when this option isTrue
we will compute the first partition of the collection to determine the metadata. This obviously triggers a compute and can take some time depending on the task graph.
Optimization specific table¶
These optimization table parameters are configured under the
awkward.optimization
namespace in the Dask configuration format.
For example, they can be set with the form:
with dask.config.set({"awkward.optimization.<option>": value}):
...
enabled
(default:True
): Enable dask-awkward specific optimizations. More fine tuning can be handled with thewhich
option.which
(default:[columns, layer-chains]
): Which of the optimizations to run. The default setting is to run all available optimizations. (ifenabled
is set toFalse
this option is ignored).columns-opt-formats
(default:[parquet]
): Which input formats should use the column optimization. The posibilities areparquet
(on by default) andjson
(currently opt-in). More information can be found in the necessary columns optimization section of the docs.on-fail
(default:warn
): When set towarn
throw a warning of the optimization fails and continue without performing the optimization. If set toraise
, raise an exception at optimization time. If set topass
, silently skip the optimization. More information can be found in the necessary columns optimization section of the docs.