NEWS.md
In publicdataau: Australian Government Open Data from 'publicdata.au'

publicdataau 0.4.1

pd_read() stops with a clear error when 'arrow' was built without zstd, which the site's Parquet files use, and its example is skipped there.

Answers are tibbles. pd_rows() and pd_aggregate() type each column as its field says, so dates come back as Date and booleans as logical, and every answer labels its columns with the fields' descriptions.
pd_fields() lists a dataset's fields with their types, descriptions, ranges and values.
pd_join_boundaries() attaches the ABS boundary of the area each row names by its code, and pd_boundaries() and pd_boundary_layers() read the layers.
pd_catalogue() searches every dataset on the government portals, served or not.
pd_provenance() gives a version's source file, checksum and fetch time.
pd_url() gives a file's address and pd_latest() the newest version.
pd_browse() opens a dataset's page.
pd_read() takes columns to read only some fields.
Errors have the class publicdataau_error, and publicdataau_unreachable when the site cannot be reached.
Large downloads show a progress bar in interactive sessions.
A vignette, "Getting started with publicdataau", and a CITATION file.

pd_tbl() returns a table or view of a version's DuckDB file as a lazy 'dplyr' table, so dplyr verbs run as SQL against the file over HTTPS.
pd_sf() reads a dataset's map layer as an 'sf' object.
pd_read(), pd_download(), pd_connect(), pd_tbl() and pd_sf() take cache to keep a version's file in pd_cache_dir() and reuse it; nothing is kept unless asked. pd_cache_list() and pd_cache_clear() manage it.
pd_changes() lists what each release changed and pd_diff() reads one comparison in full.
pd_cite() gives a version's citation as a bibentry, for text or BibTeX.
pd_datasets() takes publisher, topic and jurisdiction.
A licence condition beyond attribution, such as G-NAF's, is shown once per session; options(publicdataau.quiet = TRUE) turns it off.
pd_connect() carries the file's attribution and citation, so pd_attribution() works on a connection, and keeps 'DuckDB''s extensions out of the home directory on 'duckdb' releases before 1.5.5 too.
pd_download() takes the "geo.parquet" format.

pd_connect() attaches a version's DuckDB file read-only over HTTPS and returns a 'DBI' connection, so a query runs against the version without a download.
pd_tables() and pd_schema() list the tables of a database such as G-NAF with their fields, keys and references.
pd_read() and pd_download() take table to read one table of a database as Parquet.
pd_download() takes the new "duckdb" format.
pd_available() says whether publicdata.au can be reached, and every function now stops with a message naming the site when it cannot.