[m-rev.] for review: "Did you mean ..." mesages for long options

[Julien Fischer]

In the absence of a review, I intend to commit this one over
the weekend.

Julien.


On Sat, 6 Apr 2024, Julien Fischer wrote:

>
> "Did you mean ..." messages for long options.
>
> Generate "Did you mean ..." messages for unrecognized long options.
> For example:
>
>     unrecognized option `--no-half-at-wrn'.
>     (Did you mean `--no-halt-at-warn'?)
>
> compiler/mercury_compile_main.m:
>     If we have an unrecognized long option or negated long option,
>     then attempt to generate a "Did you mean ..." message for it.
>
> compiler/error_spec.m:
>     Add a version of maybe_construct_did_you_mean_pieces that lets
>     you add prefix to the strings before including them in the pieces.
>     This is required for handling the "--" and "--no-" long option
>     prefixes. Both versions of maybe_construct_did_you_mean pieces
>     are implemented using the same underlying piece of code.
>
> compiler/options.m:
>      Add predicates for returning the set of all long options and all
>      negatable
>      long options. To make this possible, shift the long option table into a
>      new predicate and add an (out, out) is multi mode for it.
>
> library/getopt.m:
>     Update the documentation to include maybe_string_special options
>     among the option types that are negatable.

...