Am 13.12.2016 um 10:17 schrieb clipka:

>> I still think the dot is superfluous.
> 
> Think about it:
> 
>     #declare Foo = "Dang";
>     #declare Bar = "What";
> 
>     #declare Fnord = dictionary {
>       Foo: 42,
>       Bar: sphere { <0,0,0>, 1 }
>     }
> 
> Which of the following would that supposedly be equivalent to?

To elaborate more on that point: When designing the syntax for
dictionary initializers, one could come up with the following
alternatives to all denote the same:

(Intent)

    #declare Fnord = dictionary;
    #declare Fnord["Foo"] = 42;
    #declare Fnord["Bar"] = sphere { <0,0,0>, 1 }

(A)

    #declare Fnord = dictionary {
      Foo: 42,
      Bar: sphere { <0,0,0>, 1 }
    }

(B)

    #declare Fnord = dictionary {
      "Foo": 42,
      "Bar": sphere { <0,0,0>, 1 }
    }

(C)

    #declare Fnord = dictionary {
      .Foo: 42,
      .Bar: sphere { <0,0,0>, 1 }
    }

(D)

    #declare Fnord = dictionary {
      ["Foo"]: 42,
      ["Bar"]: sphere { <0,0,0>, 1 }
    }

Note how variants (A) and (B) are fundamentally incompatible: If the key
can be a string literal, it is legitimate to also expect other string
expressions to work, and this would include plain references to string
variables. Thus, a consistent implementation of (B) would also allow for
the following:

    #declare Key1="Foo";
    #declare Key2="Bar";
    #declare Fnord = dictionary {
      Key1: 42,
      Key2: sphere { <0,0,0>, 1 }
    }

This however would take on a completely different meaning under variant (A).

Now one could rule in favour of either A or B, and I think both have
their merits. Which is why I think it wise to go for neither of them,
and instead opt for variants that are collision-free not only with
respect to each other, but even with respect to the entire set of
_possible_ alternatives.

Post a reply to this message