Releases: DataDog/sketches-py
Releases · DataDog/sketches-py
3.0.1
3.0.0
Other Notes
- This release drops support for Python versions older than 3.7.
- This change makes protobuf an optional requirement. It can be installed with
pip install ddsketch[serialization]
.
2.0.4
Bug Fixes
- Fix merging stores with max_key=0.
2.0.3
Bug Fixes
- Add support for protobuf 4.
2.0.2
Bug Fixes
- Exclude the tests module from the package.
2.0.1
Other
- Add support for protobuf>=3.0.0.
2.0.0
numpy has been removed as a dependency.
New Features
- Add support for Python 2.
- Add support for Python 3.10.
- The implementations of stores and mappings are now exposed via the top level module
ddsketch
. - Add typing.
- The package version is now exposed through
ddsketch.__version__
.
Upgrade Notes
-
DDSketch
attributesmapping
,store
,negative_store
,zero_count
,relative_accuracy
,min
andmax
have been removed. -
DDSketch.copy
method has been removed. -
DDSketch.count
attribute has been made read-only. -
DDSketch.mergeable
method has been removed. -
The custom exceptions
IllegalArgumentException
andUnequalSketchParametersException
as well as theddsketch.exceptions
module have been removed.IllegalArgumentException
andUnequalSketchParametersException
are replaced withValueError
. -
BaseDDSketch.get_quantile_value
will now returnNone
instead ofnumpy.NaN
if the specified quantile is empty.
1.1.2 Removes GK
Notable changes
- GK removed
- ddsketch is moved one level down
1.0.2 PyPI release
v1.0.2 bump version
1.0
Brings sketches-py to parity with https://github.com/DataDog/sketches-java
Notable changes
- Separates out Store and KeyMapping to make it easier to use alternatives.
- Adds protobuf serialization