Glyphs for HDF5 dataspaces and data storage are shown here. These glyphs, together with the datatype glyphs will define the dataset as it is rendered in the graph display.
Here we use the same HDF5 dataset used in the datatypes description (bravo) to illustrate the different visual elements1-4of the dataspace. For the datatype, we can see that we have a nested composite—a 1-dimensional array of length 3 containing 4-member compound/structured datatypes. Now, we need the dataspace structure and shape, and info on the storage.
The Visual Vocabulary of the Dataspace
The rank of the dataspace is affixed as a suffix to the named data object, which will be either a dataset or an attribute. Zeta is used to denote the vacuous dataspace of HDF5 null datasets.
Storage strategy for the dataspace. This glyph is visually overloaded to show two variations on storage:
- A "pipeline" glyph is used for a chunked storage that has a filter pipeline applied, with the number of filters shown as a superscripted suffix, and
- A "none" glyph is used for a) named datatypes, or b) for datasets in which the dataspace is vacuous (the HDF5 null dataset).
Total number of data elements summed over the dataspace and the dimensions of the datatype. The data elements may themselves be composites—this total does not sum over descendant datatypes in nested composites. A null dataset may have a datatype which has a shape, which is indicated by the vacuous glyph.
The shape of the dataspace, including chunk dimensions if any.
Annotation to show re-sizeable dimensions of a dataspace, either limited (single asterisk) or unlimited (double asterisk).
