IMPROVEMENTS

• new method .hdf5.isThreadsafe (returns bool) to indicate whether hdf5 lib has multi threading enabled during its build
• concurrency/threading details added to README.md
• prealloc memory rather than appending for reading (compound data/compound string)
• linux/win releases now built with 1.10.9 (users can rebuild with own version of choosing if not using this version or want multithread version using src attached to release)
• mac releases now built with hdf5 installed via brew install [email protected]
• can use env variables Q_SCRIPT_DIR / Q_SHARED_LIB_DIR to override install locations (using install.sh or install.bat)
• .hdf5.ls returns a dict rather than printing to console (ref: issue #35 )

FIX

• fix memory leak reading compound data
• changed funcs to open files in read only (rather than read write) to aid multithreaded use/etc

.hdf5.fileSize
.hdf5.dataSize
.hdf5.isAttr
.hdf5.isObject
.hdf5.getDataShape
.hdf5.getAttrShape
.hdf5.readData
.hdf5.readAttrDataset
.hdf5.copyObject (src file)

• fixed some broken links in readme due to hdf5 site changes

Promoting 1.0.0-rc.2 to 1.0.0

• Fixed Length strings both in simple and compound had been initialising index at 1 rather than 0. As such arrays were being 'stepped over'
• CMakeLists.txt file change required to remove bug that sometimes required cmake to be applied twice

• Addition of support for and release of Windows 64-bit.
• Update to the travis.yml file to standardise with Solace/MQTT/prometheus-exporter
• *.q scripts moved from root directory to q/
• Addition of test.q file to run kdb+ tests of hdf5 functionality

Hierarchical Data Format (HDF), is a widely used file format, designed to store and organize large amounts of data.
HDF5 is used extensively within the scientific and data-science/ML communities.
The Kx HDF5 interface enables conversion of q objects to/from HDF5, along with utility functions for interrogation and modifications of HDF5 datasets.