Releases: TileDB-Inc/TileDB-R
0.10.2
Release 0.10.2 of the TileDB R package supports TileDB Embedded 2.5.3 (or earlier, as well as the dev version).
Entries from the NEWS.md file follow:
tiledb 0.10.2
- This release of the R package builds against TileDB 2.5.3, but has been tested against previous releases, and the development version.
Improvements
-
The
stopifnot()assertions now use consistent error messages across all functions (#331) -
A helper function matching TileDB data types to R types is now exported (#336)
Bug Fixes
0.10.1
Release 0.10.1 of the TileDB R packages supports TileDB Embedded 2.5.2 (or earlier, as well as the dev version).
Entries from the NEWS.md file follow:
tiledb 0.10.1
- This release of the R package builds against TileDB 2.5.2, but has been tested against previous releases, and the development version.
Improvements
- An accessor for the most-recent error message string has been added (#327)
Bug Fixes
- On Linux, if a pre-made TileDB Embedded library is used, lack of AVX2 instructions is now detected and a suitable build is deployed (#328)
0.10.0
Release 0.10.0 of the TileDB R packages supports TileDB Embedded 2.5.1 (or earlier, as well as the dev version).
Entries from the NEWS.md file follow:
tiledb 0.10.0
- This release of the R package builds against TileDB 2.5.1, but has been tested against previous releases, and the development version.
Improvements
-
CI tests were expanded to also test refactored TileDB Embedded readers (#310), and now deactivated as this is now part of release 2.5.0 (#321)
-
The minimal version of TileDB Embedded that can be used with the R package is now release 2.0.0 (#313)
-
The package now compiles using the C++17 standard just like TileDB Embedded (#314)
-
Shared-memory interprocess communication is used to accelerate operation for TileDB Cloud (#316)
-
The long-deprecated
max_element_sizefunction has been removed from TileDB Embedded, and the R interface was updated accordingly (#317, #319) -
The
extendedtoggle and field fortiledb_array()can now select dimension-less returns from sparse arrays (#318) -
Use of TileDB Embedded was upgraded to release 2.5.0 (#321) following earlier upgrades to 2.4.2 and 2.4.3 (#308, #312)
-
A new quickstart example using the 'memory filesystem' was added, and one another example updated (#323)
Bug Fixes
0.9.7
Release 0.9.7 of the TileDB R package supports TileDB Embedded 2.4.0 (or earlier, or later; tested with dev, 2.3.4, 2.2.9, 2.1.6 and 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.7
- This release of the R package builds against TileDB 2.4.0, but has been tested against previous releases and the development version.
Improvements
-
Accessors for context and query statistics were added (returning easily parseable JSON strings) (#293).
-
Initial support for schema evolution was added to add or drop attributes (#294).
-
Use of TileDB Embedded was upgraded to release 2.4.0 (#295)
-
Windows builds under GitHub Actions now also include the newer UCRT variant (#296).
-
The internal memory allocation has been switched to
set_{data,offset,validity}_bufferfunctions (#297). -
A convenience accessor for schema information retrieving 'dimension' or 'attribute' status has been added (#299).
-
The default array type
fromDataFramehas been change to sparse to match some optimizations in TileDB Embedded, some unit tests have been updated accordingly (#300). -
TileDB arrays can now be queried in expression using pipes (for row-wise filtering and colunb-wise selection) (#301).
-
When matrices as well as sparse matrices are written to arrays, optional row and column names are now supported as well (#303, #304).
-
The configure script was update to the standards of autoconf 2.69 as requested by CRAN (#305).
Bug Fixes
-
Use of
set_{data,offset,validity}_bufferis made conditional on TileDB 2.4.0 or later to continue builds on older versions (#298). -
Tests of piped expressions have been rewritten to be compatible with R versions earlier than 4.1.0 (#302).
-
A dangling documentation link in README.md was corrected, and another removed (#306).
0.9.6
Release 0.9.6 of of the TileDB R package supportTileDB Embedded version 2.3.4 (or earlier or later: tested with dev, 2.2.9, 2.1.6, 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.6
- This release of the R package builds against TileDB 2.3.4, but has been tested against previous releases and the development version.
Improvements
-
When retrieving results via the
[operator, incomplete queries generate a warning (#283) -
The interface to query element size of queries was extended (#282)
-
If query ends as 'incomplete', a warning is now issue (#283)
-
The status of the preceding query can now be accessed also when using a higher-level wrapper (#285)
-
Fragment Information can be accessed via high-level accessor functions (#286)
-
A preference for data type as which TileDB array data is returned can be set (#288)
-
Continuous Integration will use increased test coverage by installing more optional package (#289)
-
Use of TileDB Embedded was upgraded to release 2.3.4 (#290)
Bug Fixes
0.9.5
Release 0.9.5 of the TileDB R package with support for the TileDB Embedded library version 2.3.3 (or earlier or later: tested with dev, 2.2.9, 2.1.6, 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.5
- This release of the R package builds against TileDB 2.3.3, but has also been tested against previous releases and the development version.
Improvements
-
A query condition parser was added for use with standard (non-quoted) R expressions (#267)
-
Windows UCRT builds at CRAN are now supported (#268)
-
Use of TileDB Embedded was upgraded to release 2.3.2 (#270), and again to 2.3.3 (#280)
-
The vacuum and consolidation helper functions now use time stamp support (#271)
-
The time-travel array opening support was updated to start and end timestamps (#272)
-
Tests for both vacuuming and consolidation 'time traveling' were added (#273)
-
Nullable string string attribute support was improved (#274)
-
Ascii columns attribute support was added (#276)
-
The query parser heuristic was improved to cover ascii strings (#277)
-
Array opening uses improvemed array state consideration skippingg re-openings for better performance (#279)
Bug Fixes
-
Domain size information gathering has been corrected for uint32, uint64, and int64 attribute domains (#266)
-
Timesteps for time-traveling unit tests were adjusted to not trip up macOS continuous integration tests (#275)
-
String array buffer size calculation was corrected also allowing for all-string arrays (#278)
0.9.4
Release 0.9.4 of the TileDB R package with support for the TileDB Embedded library version 2.3.1 (or earlier or later: tested with 2.3.0, 2.2.9, 2.1.6, 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.4
- This release of the R package builds against TileDB 2.3.1, but has also been tested against previous releases and the development version.
Improvements
-
The build defaults to TileDB Embedded 2.3.* (unless another version is found during build, or explicitly selected) (#258, #264)
-
Query condition support is available for TileDB 2.3.0 or later, allowing (possibly multiple) numerical constraints on attributes (#261)
-
Multi-dimensional arrays can now be returned from (dense) arrays via a new option (#263)
-
The package is now natively supported on Arm64 ("M1") macOS system (#264)
Bug Fixes
- Dense arrays with more than two dimensions can now be written (#260)
0.9.3
A new release 0.9.3 of the TileDB R package with support for the TileDB Embedded library version 2.2.9 (or earlier or later: tested with 2.3.0, 2.1.6, 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.3
- This release of the R package builds against TileDB 2.2.9, but has also been tested against previous releases as well as the development version.
Improvements
-
Continuous integration at GitHub is now faster as suggested packages are no longer installed (#250)
-
Arrays can now be written incrementally via the higher-level replacement function
arr[] <- obj(#251) -
The default column layout for arrays written via
fromDataFrameis now column-order (#254)
Bug Fixes
0.9.2
A new release 0.9.2 of the TileDB R package with support for the TileDB Embedded library version 2.2.9 (or earlier or later: tested with 2.3.0, 2.1.6, 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.2
- This release of the R package builds against TileDB 2.2.9, but has also been tested against previous releases as well as the development version.
Improvements
-
Matrix objects can now be returned under range selections (#247)
-
Matrix return get turned on and off with setter / getter functions (#248)
Bug Fixes
0.9.1
A new release 0.9.1 of the TileDB R package with support for the TileDB Embedded library version 2.2.9 (or earlier or later: tested with 2.3.0, 2.1.6, 2.0.8)
Entries from the NEWS.md file follow:
tiledb 0.9.1
- This release of the R package builds against TileDB 2.2.9, but has also been tested against previous releases as well as the development version.
Improvements
-
A new vignette show use of TileDB array via RMariaDB and the MyTile extension to MariaDB (#221)
-
Matrices can now be returned directly from suitable two-dimensional TileDB arrays (#225)
-
More data types are supported in the non-empty domain accessor function (#229)
-
The DESCRIPTION, README.md and pkgdown site were updated (#230)
-
Creation of TileDB arrays from data.frame object has been made more robust (#238)
-
On startup, versions numbers of the R package and the TileDB Embedded library are displayed (#239)
-
The pkgdown website now shows the ChangeLog derived from this NEWS file (#240)
Bug Fixes
-
Two tests with datetime comparisons which fail only on one macOS system are now conditional (#216)
-
Result sets with all-character column now fall back to estimated result sizes (#217)
-
Setup of support for duplicate values in sparse arrays has been corrected (#223)
-
Error messages concerning an array types and selection mismatch are now clearer (#224)
-
Writes from data.frame objects to dense array revert back to column-major order (#226)
-
Tests of sparse writes to dense matrices now use UNORDERED layout (#228)
-
Data.frame returns of selected columns now coversion dimensions as well (#231)
-
Schema creation has been generalized and made more robust (#232)
-
Selection of dimension ranges now maps date and datetime values correctly (#233, #241)
-
Selection and setting of dimension ranges has been generalized and made more robust (#235, #236)