On Sunday, 13 July 2014 at 12:33:21 UTC, Timon Gehr wrote:
Yes I know and likely would have preferred that approach if we
were
speaking about designing brand new language. But right now we
already
have string mixins
They are not a great liability. They are a simple feature. They
take up less than 200 lines of code in my D frontend
implementation, _together_ with template mixins.
I am not saying it is insanely complicated but that it is not
orthogonal to existing features. And having many ways to do the
same thing is rather bad, especially if it is such high-level
abstraction.