R-based metadata accessors have been extended to also support tiledb_array arrays (#169)
configure now also checks the hardware platform before attempting a download of a prebuilt library (#170)
SystemRequirements: in the DESCRIPTION file has been expanded (#170)
A new function limitTileDBCores() controls resource use, it is being used in tests (#139)
The function tiledb_get_context() is now exported (#140)
A new S4 class tiledb_vfs provides access to the virtual file system functionality (#140)
Functionality of selected_ranges() was extended (#142)
More (signed and unsigned) integer types are supported as dimension types in sparse arrays (#143), as well as in dense arrays (#144) and as attributes (#144)
A new S4 class tiledb_query offers access to the query object functionality in the underlying library (#145, #161)
Examples are running with a lowered thread count setting per CRAN Policies (#152)
External pointer objects now use explicitly set finalizers (#149)
Users can explicitly select a TileDB Embedded shared library built to be used (#151)
Compile-time configuration was refactored and changes (#158)
Windows builds are now possible also using TileDB Embedded build 2.0.8 (#159, #164)
Continuous integration now uses Azure for macOS and Linux (#160) and GitHub Actions for Windows (#162,#165)
Conda builds no longer call install_name_tool (#133, #146)
Downloading the prebuild library accomodates multitple targets per architecture (#150)
The number of TBB threads will only be set once (#158)
Range selection for tiledb_array objects can get/set matrices defining range (#132)
The show methods are now consistently exported and documented (#134)
TileDB is listed as copyright owner in DESCRIPTION as well (#134)
The selected_ranges method for tiledb_array types was improved, and more tests were added (#135)
C++ source code was rearranged slightly with respect to possible API deprecations in the libary (#136)
A very simple example for using TileDB Cloud from R was added (#136)
The helper scripts for the package build are now in the tools/ directory (#137)
The (optional) library download now relied on suggested R packages ‘jsonlite’ and ‘curl’ (#137)
All S4 classes are now consistently documented or aliased (#117)
If needed, the build system now builds TileDB and its required component (#118)
Data.frame support has been extended further and made more robust (#119, #123, #128)
The Description: in DESCRIPTION has been refreshed (#120)
Builds on Linux and macOS can use a pre-built TileDB library (#121, #122, #124, #127)
Copyright headers were added to source files (#125)
The pkg-config helper can be used when building from source (#126)
An introductory vignette was added (#129, #131)
Added support for heterogenous domains
Added support for string dimensions
Added support for duplicate dimension values in sparse arrays
Added support for data.frame object import and conversion to dense and sparse arrays
Added enhanced support for data.frame returns from dense array
Added support for data.frame column selection (i.e. attributes) from dense array
Added support for new filter types for md5 and sha256 checksums
Added support for Date, Datetime (i.e. POSIXct) and nanosecond dense and sparse array attributes and domains
Documentation and examples were enhanced and extended
tiledb_stats_dump() is now simpler (but needs to be enabled first as usual)
Support for nanosecond and integer64 columns was added, this is an optional feature for which the nanotime (and bit64) packages need to be installed