What about hash tables ? | Val_table of (value, value) Hashtbl.t
If you have caml closures in your values (not in Val_closure but in Val_primitive), it is maybe not appropriate to use the default hash function that would choke on them. You should rather define your own hashing function --but defining a mutually recursive `value` type and Hashtable.make application will be delicate -- or change your representation of primitives to avoid ocaml functions (by reifying them into a concrete datatype for example). On Tue, Oct 18, 2011 at 1:12 AM, Diego Olivier Fernandez Pons < [email protected]> wrote: > Caml-list, > > I have to write an interpreter for a language that has arrays explicitly > indexed by anything that can be sequential (list, range, set) > > {string} airports = {"ATL", "JFK"}; > range index = 1 .. 2; > tuple recordAirport { airport : string; id : int } > {recordAirport} otherAiports = { <"ATL", 12345>, <"JFK", 42>}; > > int myArray [airports][index] = [[1, 2], [3, 4]]; > string myArray2 [a in recordAirports][i in index] = (i < a.id) ? > "unknown" : a.airport ; > int mySliceOfArray [a in recordAirports] = sum (i in index) > myArray[a.airport][i]; > > Usually the trick in interpreter implementation is to transform everything > back to "one-dimensional" objects > - simple types > - inductive constructions for lists > - curried functions > > For instance in the book "Le langage Caml" the return type of the eval > function is > > type value = > | Val_number of int > | Val_boolean of bool > | Val_pair of value * value > | Val_nil > | Val_cons of value * value > | Val_closure of closure > | Val_primitive of value -> value > and closure = { definition: (pattern * expression) list; mutable > environnement: environnement } > and environnement == (string * value) list > > I don't see however how I am going to represent a type Val_Array given that > the indexes can be of arbitrary type and in arbitrary number. > I haven't found either how to transform the arrays into inductive types > like lists to avoid the issue. > > Any suggestions ? > > Diego Olivier > -- Caml-list mailing list. Subscription management and archives: https://sympa-roc.inria.fr/wws/info/caml-list Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
