nejucomo / dimana Goto Github PK

It would be convenient to have parsers for both a unit specification and a complete value.

These would look something like:

>>> Watt = dimana.parse_unit('KG * M^2 / SEC^3')
>>> x = Watt(3)
>>> x
3 [KG * M^2 / SEC^3]
>>> x == dimana.parse('3.000 [KG * M^2 / SEC^3]')
True

I had a unity number stored in a variable named T1, and then I did this::

>>> from decimal import Decimal
>>> D=Decimal
>>> D('0.16') * T1 - O1
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: unsupported operand type(s) for *: 'Decimal' and 'Dimensional_Units_1'

I had expected it to multiple T1 by the (dimensionless) 0.16 for me.

Allegedly there is a way to do this in dimana, using dimana.Dimensional('0.16) instead of Decimal('0.16'). So I have several requests/suggestions:

1. The initial docs for dimana should mention how to create dimensionless numbers.
2. Waitaminute, the way to create a dimensionless number is with a constructor named Dimensional? That sounds wrong.
3. Is there a reason not to promote Decimals automatically to Dimensionlesses? I guess there's a risk that a user would have a number stored in a Decimal when they intended it to be a dimensioned number, like the number of Frobs, and forcing them to explicitly put Dimensionless('0.19') would let them realize that they actually need Dimensional.get_dimension('Frobs'). But I'm skeptical—why does Dimensionless('0.19') help them notice that any better than Decimal('0.19') helps them notice?

In #1 @zooko considers it confusing that Dimensional('1.0') is unitless, and also asks why not automatically coerce Decimals to unitless Dimensionals.

The docs should clarify the type checking philosophy of dimana, which is to always require application code to be explicit, and never to coerce values.

Rather than justify to users why a value without dimensional units is still considered "dimensional", just rename the class to Dimana to reinforce the fact that it's a particular abstraction with a particular API, and to avoid different assumptions different users may have depending on their backgrounds.

The unittests only exercise dimensionless units by dint of being the result of raising a dimensional quantity to the zero-th power. It would be nice to have more tests which use dimensionless quantities, such as ratios, and zero powers.

Note: This is related to a user experience in #1.

Decimal powers should be supported. They are supported by Decimal.__pow__ so the API follows both by analogy and by having a well-defined semantics to which Dimana can delegate.

It's currently not clear from the docstrings how to create dimensionless numbers, as pointed out in the user experience issue #1.