Bikeshedding: MLang language fragment syntax

[johnwikman]

There have been a couple discussions recently about what the keyword for MLang language fragments should be, without reaching a resounding consensus. Mainly regarding whether we should use the keyword frag, or stick with the keyword lang. There are motivations for both due to how the syntax looks:

-- We are language fragments, not intended to be used on our own... but we are
-- still denoted by lang!
lang LittleFoo
  syn Expr =
  | XFoo Int

  sem handle =
  | XFoo i -> XFoo (addi i 1)
end

lang LittleBar
  syn Expr =
  | XBar Bool

  sem handle =
  | XBar b -> XBar (not b)
end



-- I am a language intended to be used directly in the code!
lang FooBar = LittleFoo + LittleBar



-- I am a language intended to be used directly in the code as well, but I am
-- also partially defined in the same way as LittleFoo and LittleBar...
lang TreeFooBar = LittleFoo + LittleBar
  syn Expr =
  | XBranch (Expr, Expr)

  sem handle =
  | XBranch (l, r) = XBranch (handle l, handle r)
end



mexpr
let fb =
  use FooBar in
  ...
in

let tfb =
  use TreeFooBar in
  ...
in

As there are motivations for both, it might also be the case that neither frag or lang is the right keyword to use. Maybe something like module could be another option, also replacing syn, sem by con, match respectively.

Hopefully I have managed to capture the key motivations brought forth in recent discussions.