On 2016-08-03 22:57, Robert burner Schadek wrote:
Well, currently you have to make that choice as developer, and there is always the BufferedLexer which should be good choice is most cases. Polymorphic design was not a goal of the project, so I think it is going to be hard to add that without sacrificing to much.
I'm not talking about a polymorphic design. I'm talking about how most functions work with ranges. They adapt depending on what type the input range is. Example, there's one lexer for forward ranges and one for input ranges. Why is that necessary? It's not necessary for functions in std.algorithm to use different names for a function for different input types.
-- /Jacob Carlborg