Equivalence involving nondimensional quantity about unitfulequivalences.jl HOT 7 CLOSED

I have never seen a “type piracy warning”, can you elaborate on that?

The first edconvert methods can be simplified as follows:

edconvert(::dimtype(Unitful.Density), x::DimensionlessQuantity, e::DensityConcentration) = x * 1u"kg/L"

But I agree that this should work with the macro. I will implement it so that @eqrelation can be used with DimensionlessQuantity.

from unitfulequivalences.jl.

#7 implements the necessary dimtype method.

One thing that is still missing for dimensionless quantities to work correctly is support for plain numbers: uconvert(u"kg/L", 5u"mm/m", DensityConcentration()) works, but uconvert(u"kg/L", 5, DensityConcentration()) does not. I will implement that as well. Note that the conversion to plain numbers already works: uconvert(NoUnits, 5u"kg/L", DensityConcentration()) Support for plain numbers is now included in #7 as well.

from unitfulequivalences.jl.

Many thanks for implementing this!

And thanks for telling me about DimensionlessQuantity. I was looking for something like this to use as you mentioned, but somehow I missed it.

As for the type piracy, the warning I get is "An imported function has been extended without using module defined typed arguments.", which shows up in julia-vscode/StaticLint.jl as "TypePiracy".

I just learned about this so I will just point to two places that explain it better: julialang docs: Avoid type piracy and discourse: Clarification on type piracy.

So I agree that my extension to dimtype is a type piracy since both types I use are defined in other modules, but the following extensions use the type DensityConcentration, which is defined in my module, but they still give me this warning:

edconvert(d::dimtype(Unitful.Density), x::Unitful.DimensionlessQuantity, e::DensityConcentration) = x * 1u"kg/l"

edconvert(d::dimtype(NoDims), x::Unitful.Density, e::DensityConcentration) = x * 1u"L/kg"

Anyways, I will try out now with the direct relation that you implemented and get back here to let you know.

from unitfulequivalences.jl.

Ah, I guess I need to wait till the pull-request is closed, no worries. And I like that you proposed to make the definition of DimensionlessQuantity in Unitful.jl consistent with the others.

from unitfulequivalences.jl.

Thanks, I know about type piracy but did not know that there are linters who warn about it.

However, the edconvert methods are not type piracy (since you own DensityConcentration), so I wonder why you get a warning for them. Maybe the linter doesn’t recognise that DensityConcentration is a type from your module, since its definition is obfuscated by the @equivalence macro? Is the warning fixed by defining the DensityConcentration struct yourself instead of using the macro?

from unitfulequivalences.jl.

Is the warning fixed by defining the DensityConcentration struct yourself instead of using the macro?

If I define it directly then the warning disappears.

from unitfulequivalences.jl.

If I define it directly then the warning disappears.

One more reason to remove the @equivalence macro 😄

from unitfulequivalences.jl.