Skip to content

HDF5-UDF 1.2
Compare
Choose a tag to compare
@lucasvr lucasvr released this 29 Jan 04:04
v1.2
3273fd6

We're happy to announce the availability of HDF5-UDF version 1.2, which comes with exciting new features over its predecessor:

  • Support for outputting string datatypes with a user-configurable size
  • Support for outputting compound datatypes, including string elements and native datatypes
  • Use 1-based indexing on the Lua API to conform with the language best practices
  • Allow reading from CSV files, as long as they are not symlinks to other objects in the filesystem and that they're placed in the same directory as the HDF5 file

The most interesting aspect of outputting compounds and strings is that it's now possible to translate CSV to HDF5 on-the-fly.

Compound datasets are described using a variation of the original command-line syntax:

name:{member:type[,member:type...]}:resolution

For instance, an array with 1000 measurements of a generic sensor could be described as:

SensorData:{timestamp:int64,value:float}:1000

String elements (as well as string datasets) use the string datatype. By default HDF5-UDF allows up to 32 characters for that datatype. That can be overridden using the (N) modifier, as in string(48), which allows string elements to have up to 48 characters.

Please refer to the updated documentation for further details on how to take the best out of this release.

Instructions on how to install the package in source code form or on Debian, Ubuntu, and Fedora derivatives are available here.

Assets 3
Loading