I've recently found this working:
const Mode = 0
var x: int | string | seq[int | string] = (
when Mode == 0:
0
elif Mode == 1:
""
elif Mode == 2:
@[0]
else:
@[""]
)
echo(x, " ", typeof(x), " ", sizeof(x))
I may be missing something, what new abilities does this grant you?
To me this looks like you determine the type at compile-time and assign a default value, after that you face the same limitation as if you had written var x = 0. Figuring out which code gets written there is just a tad harder than before.
You can write same code without type class:
const Mode = 0
var x = when Mode == 0:
0
elif Mode == 1:
""
elif Mode == 2:
@[0]
else:
@[""]
echo x
Unlike if expression, when expression doesn't require each expressions in each branches are same. I don't think type classes are useful when declaring variables.
what new abilities does this grant you?
Not needing to write:
when Mode == 0:
var someVeryLongName = 0
elif Mode == 1:
var someVeryLongName = ""
elif Mode == 2:
var someVeryLongName = @[0]
else:
var someVeryLongName = @[""]