Introduction

• What’s new in h5py 2.1
  • Dimension scales
  • Unicode strings allowed in attributes
  • Dataset size property
  • Dataset.value property is now deprecated.
  • Bug fixes
• What’s new in h5py 2.0
  • Enhancements unlikely to affect compatibility
  • Changes which may break existing code
    • Supported HDF5/Python versions
    • Group, Dataset and Datatype constructors have changed
    • Unicode is now used for object names
    • File objects must be manually closed
    • Changes to scalar slicing code
    • Array scalars now always returned when indexing a dataset
    • Reading object-like data strips special type information
    • The selections module has been removed
    • The H5Error exception class has been removed (along with h5py.h5e)
    • File .mode property is now either ‘r’ or ‘r+
    • Long-deprecated dict methods have been removed
  • Known issues
• Downloading and Building h5py
  • Getting HDF5
  • Installing on Windows
    • Requires
  • Installing on Linux/Mac OS-X
    • Requires
    • Quick installation
    • Custom installation
  • Testing
• Quick Start Guide
  • What is HDF5?
  • What is h5py?
  • Getting data into HDF5
    • Getting your data back
    • Closing the file
  • Groups & multiple objects
  • Attributes
  • Other stuff
• How do I… ?
  • Files
    • Open & close
    • Open w/context manager
    • Get filename
    • Get mode (readonly or r/w)
    • Create a file with user block
    • Access a file’s user block
    • Open a memory-resident file
  • Datasets
    • Read an entire dataset as a NumPy array
    • Read a scalar dataset