Should a parser type be a struct or class?

[Meta]

On Wednesday, 17 June 2020 at 11:50:27 UTC, Per Nordlöw wrote:
> Should a range-compliant aggregate type realizing a parser be 
> encoded as a struct or class? In dmd `Lexer` and `Parser` are 
> both classes.
>
> In general how should I reason about whether an aggregate type 
> should be encoded as a struct or class?

IMO it doesn't need to be. However, it's worth saying that range 
semantics aren't a great fit for parsers - at least that's been 
my experience. Parsers need to be able to "synchronize" to 
recover from syntax errors, which does not fit into the range API 
very well. You can probably fit it in somewhere in popFront or 
front or empty, as your implementation permits, but I find it's 
just easier to forego the range interface and implement whatever 
primitives you need; *then* you can add a range interface over 
top that models the output of the parser as a range of 
expressions, or whatever you want.