NEWS

dasklyr 0.1.2

summarise() after joins no longer errors when combining n() or n_distinct() with other aggregates on two grouping columns (Dask shuffle + pandas 3 merge). n() is a dummy-ones sum in one groupby.agg(), and two aggregates may now target the same column (mean(x) and max(x)).
dask_connect(master = "coiled", cluster_name = ...) opens or attaches a 'Coiled' cluster through Python coiled.Cluster. dask_coiled_clusters() lists account clusters. Creating a cluster defaults to n_workers = 1 (1 scheduler + 1 worker). Tests: default never creates a cluster; DASKLYR_COILED=true attaches a ready DASKLYR_COILED_NAME; DASKLYR_COILED=create is required to start a new id.
DESCRIPTION: quote 'Python' as requested after the 0.1.1 CRAN acceptance.

CRAN incoming pretest: quote software names in DESCRIPTION, drop the loopback dashboard link from README, and do not start Python from examples or CRAN tests via dask_available().

First public release: connect to local or remote Dask, copy_to() / tbl(), dplyr verbs, Parquet and CSV, ddf_* table helpers, and dask_invoke().