Expand simulator to handle more DSL programs

PR #57 adds a simulator and T2T compiler to our DSL! Much of that work was to convert DSL programs into "controls" (see Definition 3.7 in Formal Abstractions or issue #53). This has been done for programs built with Strict, RoundRobin, WeightedFair, and Fifo; however, building controls is harder for other policies because the semantics of our DSL have yet to be ironed out (for example, see issue #51).

Instead of maintaining multiple catch-all branches in the simulator, we've temporarily removed certain policies from Policy.t and parked their associated code here

(* from Policy.t in both policy.ml and policy.mli *)
type t =
  | Class of Ast.clss
  | Fifo of t list
  | RoundRobin of t list
  | Strict of t list
  | WeightedFair of (t * float) list
  | EarliestDeadline of t list
  | ShortestJobNext of t list
  | ShortestRemaining of t list
  | RateControlled of t list
  | LeakyBucket of t list * int * int
  | TokenBucket of t list * int * int
  | StopAndGo of t list * int

(* from Policy.to_string *)
  match p with
  | Class c -> c
  | Fifo lst -> sprintf "fifo%s" (join lst)
  | RoundRobin lst -> sprintf "rr%s" (join lst)
  | Strict lst -> sprintf "strict%s" (join lst)
  | WeightedFair lst -> sprintf "wfq%s" (join_weighted lst)
  | EarliestDeadline lst -> sprintf "edf%s" (join lst)
  | ShortestJobNext lst -> sprintf "sjn%s" (join lst)
  | ShortestRemaining lst -> sprintf "srtf%s" (join lst)
  | RateControlled lst -> sprintf "rcsp%s" (join lst)
  | LeakyBucket (lst, width, buffer) ->
    sprintf "leaky[%s, width = %d, buffer = %d]" (join lst) width buffer
  | TokenBucket (lst, width, buffer) ->
    sprintf "token[%s, width = %d, time = %d]" (join lst) width buffer
  | StopAndGo (lst, width) -> sprintf "stopandgo[%s, width = %d]" (join lst) width

(* from Policy.sub *)
  match p with
  | Class c -> if List.mem c cl then Class c else raise (UndeclaredClass c)
  | Var x -> sub cl st (lookup st x)
  | Fifo plst -> Fifo (sub_plst cl st plst)
  | RoundRobin plst -> RoundRobin (sub_plst cl st plst)
  | Strict plst -> Strict (sub_plst cl st plst)
  | WeightedFair wplst -> WeightedFair (sub_weighted_plst cl st wplst)
  | EarliestDeadline plst -> EarliestDeadline (sub_plst cl st plst)
  | ShortestJobNext plst -> ShortestJobNext (sub_plst cl st plst)
  | ShortestRemaining plst -> ShortestRemaining (sub_plst cl st plst)
  | RateControlled plst -> RateControlled (sub_plst cl st plst)
  | LeakyBucket (plst, n1, n2) -> LeakyBucket (sub_plst cl st plst, n1, n2)
  | TokenBucket (plst, n1, n2) -> TokenBucket (sub_plst cl st plst, n1, n2)
  | StopAndGo (plst, n) -> StopAndGo (sub_plst cl st plst, n)

The plan is to add these back to Policy when the simulator can handle them.

NOTE: none of the lexing and parsing for these policies has been altered: that is, ast.ml is unchanged

cucapra / packet-scheduling

Expand simulator to handle more DSL programs #58