[m-rev.] for possible review: tidying up reference manual

[Julien Fischer]

On Fri, 19 Jul 2019, Zoltan Somogyi wrote:

> On Fri, 19 Jul 2019 09:47:05 +0200 (CEST), "Zoltan Somogyi" <zoltan.somogyi at runbox.com> wrote:
>> For now, I will add the check for nonabstract solver types in the interface.
>
> For post-commit review by anyone.

> Don't allow solver types to be exported.
> 
> compiler/add_type.m:
>     Generate an error message if a solver type is exported.
> 
> NEWS:
>     Announce the enforcement of this old rule.
> 
> tests/invalid/any_mode.err_exp:
> tests/invalid/foreign_solver_type.err_exp:
>     Expect the new error message for exported solver types.
> 
> tests/valid/solver_type_bug.m:
>     To keep this test case valid, don't export the solver type it defines.
> diff --git a/NEWS b/NEWS
> index d994085ac..781840f2c 100644
> --- a/NEWS
> +++ b/NEWS
> @@ -33,7 +33,8 @@ Changes that may break compatibility:
>
>  * We have enabled stricter checking of type declarations and definitions:
>    all declarations and definitions of a type must agree on whether the type
> -  is a solver type or not.
> +  is a solver type or not, and solver types may not be exported from their

     and solver type definitions may not

Similarly. the error message you've added need to make it clear that it
is the definition that cannot be exported.

The rest is fine.

Julien.