How should mendeleev deal with units of properties?

[lmmentel]

Many properties available in mendeleev next to numerical values have also units of measurement. The units for most properties are documented in the data page in the docs. All units except for isotope half life are consistent across all value for a given property. This means for example that all boiling points are expressed in Kelvins and covalent radii are in picometers.

Thanks to recent changes #156 units are now programmatically available through mendeleev.models.PropertyMetadata and therefore can be used to introduce an interface where quantities are returned with explicit unit.

One popular option for modelling units explicitly would be by incorporating the pint package for handling untis.

Open design questions:

• should mendeleev adopt explicit unit treatment for dimensional properties?
• should there be 2 interfaces available, with and without explicit units?
• how would a good interface/api with explicit unit treatment look like?

Context

• previous discussion Half_life of isotopes return float, but unit differs depending on which isotopes is requested #94

[kalvdans]

Thanks for bringing up the discussion! I think mendeleev should always return SI units, but might provide a library to print a value in SI units to a human-readable string.