Version française
Home     About     Download     Resources     Contact us    

This site is updated infrequently. For up-to-date information, please visit the new OCaml website at

Browse thread
[Caml-list] recursive modules
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: 2003-05-05 (08:21)
From: Xavier Leroy <xavier.leroy@i...>
Subject: [Caml-list] recursive modules
Since the issue of recursive modules has (again) popped up on this
list, it seems that now is a good time to announce an experimental
design and implementation for recursive modules in OCaml that I've
been working on lately.  A note describing the design is at

and the implementation can be downloaded from the CVS repository
on, tag "recursive_modules".

What this extension provides is a "module rec ... and ..." binding
that allows the definition of mutually-recursive modules within the
same compilation units.  Recursion between compilation units is a
different problem that is not adressed yet.  To give a flavor of the
extension, one can write for instance

    module A : sig
                 type t = Leaf of string | Node of ASet.t
                 val compare: t -> t -> int
             = struct
                 type t = Leaf of string | Node of ASet.t
                 let compare t1 t2 =
                   match (t1, t2) with
                     (Leaf s1, Leaf s2) -> s1 s2
                   | (Leaf _, Node _) -> 1
                   | (Node _, Leaf _) -> -1
                   | (Node n1, Node n2) -> n1 n2
    and ASet : Set.S with type elt = A.t
             = Set.Make(A)

The other point worth mentioning is that the detection of ill-founded
recursive definitions (definitions that have no fixpoint in a
call-by-value evaluation regime) is not completely static and may
cause an "Undefined" exception to be thrown at program initialization
time.  Fully static prevention of ill-founded recursion would, in the
current state of our knowledge, either rule out too many valuable
uses, or require major complications in the type system.  The proposed
approach is a pragmatic compromise rather than a 100% intellectually
satisfying solution.

No decision has been taken yet concerning the possible integration of
this extension in a future release of OCaml.

Comments and feedback are most welcome, as long as they are of the
constructive kind.

- Xavier Leroy

To unsubscribe, mail Archives:
Bug reports: FAQ:
Beginner's list: