leaky abstraction in makebaseunit

[mfey]

(makebaseunit <string>) is currently just a layer of syntactic sugar over JScience's BaseUnit.

If we try to use the name of a unit already defined in JScience's Unit/SI as the name of a new dimension, we get an exception. This is clearly a bug.

At the repl:

user=> (require '[units2.IFnUnit :refer [defbaseunit]])
nil
user=> (defbaseunit s "G")
#'user/s
user=> (defbaseunit s "H") ; Henry
CompilerException java.lang.IllegalArgumentException: Symbol H is associated to a different unit, compiling:(form-init######.clj:1:1) 
user=> (defbaseunit s "I")
#'user/s
user=> (defbaseunit s "J") ; Joule
CompilerException java.lang.IllegalArgumentException: Symbol J is associated to a different unit, compiling:(form-init######.clj:1:1)
user=> (defbaseunit s "K") ; Kelvin, so expect exception, but it seems to work?
#'user/s

Intriguingly, using units from Unit/NonSI is just fine:

user=> (defbaseunit s "a") ; are
#'user/s
user=> (defbaseunit s "b")
#'user/s
user=> (defbaseunit s "c") ; speed of light
#'user/s
user=> (defbaseunit s "d") ; day
#'user/s
user=> (defbaseunit s "e") ; positron charge
#'user/s

An exception to this exceptional behaviour (!) occurs for Baseunits defined in Unit/SI themselves ("A", "cd", "K", "kg", "m", "mol", "s"), which seem to work just fine at a first glance but actually don't do what you'd expect!

user=> (defbaseunit s "A") ; ampere -> intensity
#'user/s
user=> (.getDimension s)
#object[javax.measure.unit.Dimension 0x29559cae "[I]"]

user=> (defbaseunit s "cd")
#'user/s
user=> (.getDimension s)
#object[javax.measure.unit.Dimension 0x22759915 "[L]²·[M]/[T]³"]

user=> (defbaseunit s "K")
#'user/s
user=> (.getDimension s)
#object[javax.measure.unit.Dimension 0x724b8f81 "[θ]"]

user=> (defbaseunit s "s") ; second -> time
#'user/s
user=> (.getDimension s)
#object[javax.measure.unit.Dimension 0x2b1781ab "[T]"]