Common Error Messages

This page gives a list of quick explanations for some error or warning messages that are emitted by the OCaml compilers. Longer explanations are usually given in dedicated sections of this tutorial.

Type Errors

This expression has type ... but is here used with type ...

When the type of an object is not compatible with the context in which it is used, it is frequent to obtain this kind of message:

# 1 + 2.5;;
Line 1, characters 5-8:
Error: This expression has type float but an expression was expected of type
         int

"This expression has type X but is here used with type Y" means that if the contents of the expression is isolated (2.5), its type is inferred as X (float). But the context, i.e. everything which is around (1 + ...) tells that the gap expects an expression of type Y (int) which is not compatible with X.

More disturbing is the following message:

This expression has type my_type but is here used with type my_type

This error happens often while testing some type definitions using the interactive toplevel. In OCaml, it is perfectly legal to define a type with a name that is already taken by another type. Consider the following session:

# type my_type = A | B;;
type my_type = A | B
# let a = A;;
val a : my_type = A
# type my_type = A | B;;
type my_type = A | B
# let b = B;;
val b : my_type = B
# a = b;;
Line 1, characters 5-6:
Error: This expression has type my_type/1
       but an expression was expected of type my_type/2
       Hint: The type my_type has been defined multiple times in this
         toplevel session. Some toplevel values still refer to old versions
         of this type. Did you try to redefine them?

For the compiler, the second definition of my_type is totally independent from the first definition. So we have defined two types which have the same name. Since "a" was defined earlier, it belongs to the first type while "b" belongs to the second type. In this example, redefining "a" after the last definition of my_type solves the problem. This kind of problem should not happen in real programs unless you use the same name for the same type in the same module, which is highly discouraged.

Warning: This optional argument cannot be erased

Functions with optional arguments must have at least one non-labelled argument. For instance, this is not OK:

# let f ?(x = 0) ?(y = 0) = print_int (x + y);;
Line 1, characters 18-23:
Warning 16 [unerasable-optional-argument]: this optional argument cannot be erased.
Line 1, characters 9-14:
Warning 16 [unerasable-optional-argument]: this optional argument cannot be erased.
val f : ?x:int -> ?y:int -> unit = <fun>

The solution is simply to add one argument of type unit, like this:

# let f ?(x = 0) ?(y = 0) () = print_int (x + y);;
val f : ?x:int -> ?y:int -> unit -> unit = <fun>

See the Labels section for more details on functions with labelled arguments.

The type of this expression... contains type variables that cannot be generalized

This happens in some cases when the full type of an object is not known by the compiler when it reaches the end of the compilation unit (file) but for some reason it cannot remain polymorphic. Example:

# let x = ref None;;
val x : '_weak1 option ref = {contents = None}

triggers the following message during the compilation:

The type of this expression, '_a option ref,
contains type variables that cannot be generalized

Solution: help the compiler with a type annotation, like for instance:

# let x : string option ref = ref None;;
val x : string option ref = {contents = None}

or:

# let x = ref (None : string option);;
val x : string option ref = {contents = None}

Data of type '_weak<n> may be allowed temporarily, for instance during a toplevel session. It means that the given object has an unknown type, but it cannot be any type: it is not polymorphic data. In the toplevel, our example gives these results:

# let x = ref None;;
val x : '_weak2 option ref = {contents = None}

The compiler tells us that the type of x is not fully known yet. But by using x later, the compiler can infer the type of x:

# x := Some 0;;
- : unit = ()

Now x has a known type:

# x;;
- : int option ref = {contents = Some 0}

Pattern Matching Warnings and Errors

This pattern is unused

This warning should be considered as an error, since there is no reason to intentionally keep such code. It may happen when the programmer introduced a catch-all pattern unintentionally such as in the following situation:

# let test_member x tup =
  match tup with
  | (y, _) | (_, y) when y = x -> true
  | _ -> false;;
Line 3, characters 14-20:
Warning 12 [redundant-subpat]: this sub-pattern is unused.
Line 3, characters 5-20:
Warning 57 [ambiguous-var-in-pattern-guard]: Ambiguous or-pattern variables under guard;
variable y appears in different places in different or-pattern alternatives.
Only the first match will be used to evaluate the guard expression.
(See manual section 11.5)
val test_member : 'a -> 'a * 'a -> bool = <fun>

Obviously, the programmer had a misconception of what OCaml's pattern matching is about. Remember the following:

  • the tree of cases is traversed linearly, from left to right. There is no backtracking as in regexp matching.
  • a guard ("when" clause) is not part of a pattern. It is simply a condition which is evaluated at most once and is used as a last resort to jump to the next match case.
  • lowercase identifiers (bindings such as "y" above) are just names, so they will always match.

In our example, it is now clear that only the first item of the pair will ever be tested. This leads to the following results:

# test_member 1 (1, 0);;
- : bool = true
# test_member 1 (0, 1);;
- : bool = false

This pattern-matching is not exhaustive

OCaml's pattern matching can check whether a set of patterns is exhaustive or not, based on the type only. So in the following example, the compiler doesn't know what range of ints the "mod" operator would return:

let is_even x =
  match x mod 2 with
  | 0 -> true
  | 1 | -1 -> false

A short solution without pattern matching would be:

# let is_even x = x mod 2 = 0;;
val is_even : int -> bool = <fun>

In general, that kind of simplification is not possible and the best solution is to add a catch-all case which should never be reached:

# let is_even x =
  match x mod 2 with
  | 0 -> true
  | 1 | -1 -> false
  | _ -> assert false;;
val is_even : int -> bool = <fun>

Problems Recompiling Valid Programs

x.cmi is not a compiled interface

When recompiling some old program or compiling a program from an external source that was not cleaned properly, it is possible to get this error message:

some_module.cmi is not a compiled interface

It means that some_module.cmi is not valid according to the current version of the OCaml compiler. Most of the time, removing the old compiled files (*.cmi, *.cmo, *.cmx, ...) and recompiling is sufficient to solve this problem.

Warning: Illegal backslash escape in string

Recent versions of OCaml warn you against unprotected backslashes in strings since they should be doubled. Such a message may be displayed when compiling an older program, and can be turned off with the -w x option.

# "\e\n" (* bad practice *);;
File "_none_", line 1, characters 1-3:
Warning 14 [illegal-backslash]: illegal backslash escape in string.
- : string = "\\e\n"
# "\\e\n" (* good practice *);;
- : string = "\\e\n"

Help Improve Our Documentation

All OCaml docs are open source. See something that's wrong or unclear? Submit a pull request.

OCaml

Innovation. Community. Security.