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] limits on mutual recursion and modules?
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: -- (:)
From: Brian Rogoff <bpr@b...>
Subject: Re: [Caml-list] limits on mutual recursion and modules?
On Wed, 21 Nov 2001, William Harold Newman wrote:
> I'm trying to understand the limits on mutual recursion in ML.
> I've seen the hack
>   type 'a combination = T1 of int | T2 of 'a | T3 of 'a * 'a
>   class virtual test =
>     object
>       method virtual get: test combination
>     end
> for mutual recursion between classes and modules in OCaml. That doesn't 

You mean between classes and types here, of course. 

> leave me with much confidence that I can figure out whether a particular
> kind of mutual recursion is possible.:-|
> I've seen various statements about recursion between modules being
> impossible, but I'm not sure exactly how severe a limitation this is
> in practice, especially given the possibility of hacks like the one
> above.

That hack, which I've seen called the "parameterization trick" (we really
need a better, sexier sounding name for it) is the way you currently 
create a recursion between a type definition and a functor instantiation.

> In particular, I'm curious whether it's possible to define
> a record type Foo which contains a functor-defined data structure which 
> refer to objects of type Foo. E.g., in OCaml is there any way
> to define a record type Foo one of whose fields is a Set of Foo?

You use that same trick. It also means that you must make a polymorphic
version of Set to participate in the recursion; the library Set won't do. 
Check this out

It's my guess that every frequent user of OCaml or SML bangs into this
within their first 9 months of serious ML programming, and most likely
long before that.

> In general, I'd be interested in any pointers to treatments of this
> problem and the theoretical limits involved.

as he is the one working on it and he has pointers to related work there. 

-- Brian
Bug reports:  FAQ:
To unsubscribe, mail  Archives: