What are the names of types?

[willow-ahrens]

When the descriptor says "uint64", I think I know how to make the appropriate type, but how can I resolve these type names more formally? Are they names of C datatypes?

[eriknw]

Ah, right, the types ought be explicit in the spec.

The integer type names are from stdint.h C header file:

int8: int8_t
int16: int16_t
int32: int32_t
int64: int64_t
uint8: uint8_t
uint16: uint16_t
uint32: uint32_t
uint64: uint64_t

And floating point types are:

float32: float or float32_t (IEEE 754 binary32)
float64: double or float64_t (IEEE 754 binary64)

I don't know if we have other types such as bool, complex64, int128, float16, strings (constant or variable size), datetimes, timedeltas, and user-defined types pinned down yet. I believe we want to defer to the underlying storage to handle endianness.

Some possibly useful links:

• https://arrow.apache.org/docs/cpp/api/datatype.html
• https://numpy.org/doc/stable/user/basics.types.html
• http://davis.lbl.gov/Manuals/HDF5-1.8.7/UG/11_Datatypes.html
• https://zarr.readthedocs.io/en/stable/spec/v2.html#data-type-encoding

[eriknw]

@jim22k and @BenBrock have been closer to implementations and can probably say where we are with e.g. bool.

[willow-ahrens]

I see! All of these names are sortof up to the implementer to represent faithfully in the host language, so I think that uint8_t would be a fine choice for the hdf5 array type, and "bool" would be a fine string to hint for the user to decide how bools get represented in their language.

[willow-ahrens]

I'll update my parser, and I'd be happy to add a PR for the spec for this if you'd like.