- Bugfix:
DataBackendDuckDB
calculated missing values incorrectly. - Compatibility with future versions of
dbplyr
(#35).
- Compatibility with new duckdb version (#36).
- Support for parquet files as Backend via DuckDB.
- New converter
as_duckdb_backend()
.
- Compatibility fixes with new duckdb version.
- Temporarily disabled some tests to overcome some regressions in duckdb.
- Added a
show_query()
method forDataBackendDplyr
(#4). - A reconnector is automatically added in
as_data_backend()
for objects of typetbl_SQLiteConnection
andtbl_duckdb_connection
.
- New backend
DataBackendDuckDB
. dplyr
is now optional (moved from imports to suggests).
- Set a primary key for SQLite databases generated from data frames.
- Set a reconnector for SQLite databases generated from data frames.
- Resolved a warning signaled by dplyr-1.0.0.
as_data_backend()
method to construct aDataBackendDplyr
now specialized to operate on objects of type"tbl_lazy"
(was"tbl"
before). This way, local"tbl"
objects such as tibbles are converted to aDataBackendDataTable
bymlr3::as_data_backend.data.frame()
.
- Connections can now be automatically re-connected via a user-provided function.
DataBackendDplyr
now has a finalizer which automatically disconnects the database connection during garbage collection.
- During construction of
DataBackendDplyr
, you can now select columns to be converted from string to factor. This simplifies the work with SQL databases which do not naturally support factors (or where the level information is lost in the transaction).
- Fixed
$distinct()
to not return missing values per default. - Added
na_rm
argument to$distinct()
. - Renamed
as_sqlite()
toas_sqlite_backend()
- Initial release.