Releases: bsc-dd/hecuba
Releases · bsc-dd/hecuba
Hecuba 2.1
C++
- Allow creation of StorageNumpy with different types (float, boolean and byte)
Hecuba 2.0
C++ interface
- Simplify user interface
- Remove yaml dependency
- Add Extrae instrumentation
Python
- Upgrade to 3.10
Documentation
- Update documentation for C++ interface
- Update README
- Add missing documentation on Python requirements
- Add documentation for c4s
ARROW
- Simplify deployment making ARROW optional: Add USE_ARROW compilation flag
StorageNumpy
- Add HECUBA_LAZY_WRITES to enable/disable lazy writes
Performance improvements
- Improve SYNC operation time.
- Improve-writer
- Use a single thread for write operations and a single table for
StorageNumpys (HECUBA_SN_SINGLE_TABLE): new-writer-thread
Create a static CacheTable to access StorageNumpys (for single table case)
C4S
- Allow to configure number of cores fo application and cassandra
Bug Fixing
- Fix an installation problem: fix-missing-storage-jar
- Fix an installation problem with missing '--prefix'
- Fix python arrow installation
- Fix install process
- Fix bug when processing attributes read from cassandra: fix bug in IStorage::extractMultiValuesFromQueryResult
- Fix a bug in IStorage::setCache
- Fix writes of a single column into cassandra
- Fix missing return value in attribute
- Add thread support to alive objects state
- Fix bug when disconnecting Cassandra
- Make sure Cassandra is available
Hecuba 1.2.4
Add missing python support for StorageStream class
Hecuba 1.2.3
Changelog
C++ interface
- Add C++ API
- Add support to write a single attribute in storageobj.
Stream interface
- Streaming tools deployment
- Add streaming API
API:
- Add 'sync' to API
- Modify split with optional parameter (cols=true/false)
Others:
- CREATE_SCHEMA to default true
- Change UUID generation (UUID3->UUID5)
Bugfixing
Hecuba 1.0
Changelog
- Major code refactoring and code optimization
- Improve ndarray management and compatibility with numpy library
- Add a synchronization interface
- Add support to concurrent creation
- Add experimental support to Arrow format (disabled by default)
- Improve automatic deployment on slurm based systems
Hecuba 0.1.3 Python 3
Change log
- Upgrade to Python 3
- Automated tests for Python versions 3.6 & up
- Tuples support None as one of their values #248
- Bug fixes
Hecuba 0.1.2 Python2.7
Change log
- New cache with low overhead #97 #234
- Timestamped writes to ensure serial updates #233
- StorageDict now supports deletes by key (del dict[key]) with coherency #98 #184
- Support for "None" Python objects, which are translated to "null" in Cassandra. #107 #170
- Persistent nested objects fully supported (e.g StorageDict of StorageObj) #123 #126 #135 #139 #190
- Travis automation and integration, building wheels too #129 #247
- Codecoverage integration #247
- Multinode testing: Python tests include their own Cassandra cluster with CCM. #243
- Changes in StorageNumpy are persisted, and basic operations supported #134
- Method
split
now takes into account Cassandra Virtual nodes for data locality #243 - Update integration with Qbeast #226 #231
- Enhanced filter on persistent objects which now adheres to Python's syntax #221
- Update CQL syntax to be compatible with ScyllaDB #174
- Add support for tuple data types #222
- Add support for sets (with restrictions) #203 #211
- Cassandra C++ Driver can be tuned with environment variables #182
- Bug fixes
Updated PyCOMPSs compability
Change log
- StorageDict returns KeyError when key not found
- Fixed StorageItf implementation
- Fixed dependency loop in storage.api.getByID
- Updated tests to compensate cache incoherence
Tempting alpha release
Dropped unstable features as numpy support and nested storage objects.