[Ocaml-lib-devel] unknown length, user defined enumerations?

[Brian H. <bri...@ql...>]

Something just occured to me, and I don't know if this is a good idea or 
not.  But we were talking about reading a file and tail recursion over in 
Ocaml beginners just now.  And I whipped up a quick example of reading a 
file into a list of strings, using the standard List.rev trick.

Now, that's "correct" for Ocaml beginners, but not what you really want to 
do.  What you really want to do is use setcdr to construct the list 
forwards, instead of backwards.  But setcdr is dangerous, and not 
something I want to blithely recommend beginners play around with.

What I thought would be neat would be to be able to allow the user to 
define their own enumeration.  Define a function like:

    let next chan () = (* partially apply to define chan *)
        try
            input_line chan
        with
            End_of_file -> raise No_more_elements
    ;;

You then pass this next function to list_of_enum and it uses setcdr in a 
safe way to create your list of the lines of a file.  This allows us to 
have safe setcdr-based list building in a generic way- *without* having to 
discuss the dangers of setcdr with beginners.  Enumerations are a simple 
enough idea- Java and C++ both have their iterators.

Here's the problem: you don't know, ahead of time, how many lines are in 
the file.  So there's no sane way to implement a count function.  The 
count function, if meaningful, can be usefull- for example, Xarray could 
make sure only one reallocation is necessary to add an entire enumeration.  
But it's not *necessary*.

So I propose that count be allowed to return -1 to indicate "length 
unknown".  Throwing an exception might also be an idea.  Enum users who 
depend upon count should be prepared to handle unknown length enumerations 
(possibly at reduced efficiency- Xarray, for example, might need to 
reallocate multiple times instead of just once).

This would allow you to read an entire file into a list using setcdr just 
by

let readfile chan =
    let next () = 
        try
            input_line chan
        with
            End_of_file -> raise No_more_elements
    and count () = -1 (* or raise Length_unknown *)
    in
    List.of_enum (Enum.make ~next:next ~count:count)
    ;;

Thoughts?  Comments?

Brian