[m-rev.] for post-commit review: --show-movability

[Julien Fischer]

On Sun, 1 Oct 2023, Zoltan Somogyi wrote:

> I am seeking opinions on the name of the new option,
> which generates reports such as the ones attached,
> and on its documentation. Once we have consensus
> on that, I will update doc/reference_manual.texi,
> and also NEWS.md if people think this is a good idea.

I think that is a good idea.

> Add a new option --show-movability.

Perhaps, --show-pred-movability.

...

> diff --git a/compiler/hlds_call_tree.m b/compiler/hlds_call_tree.m
> index 03d360e43..170333c05 100644
> --- a/compiler/hlds_call_tree.m
> +++ b/compiler/hlds_call_tree.m

...

> +    ( if set_tree234.is_empty(ConflictExportedPredIdSet) then
> +        ConflictExportedPieces = []
> +    else
> +        ConflictExportedPredPieces = name_set_to_line_pieces(ModuleInfo,
> +            ConflictExportedPredIdSet),
> +        ConflictExportedPieces =
> +            [words("Moving these predicates and/or functions to a new module"),
> +            words("would require the new module to import the current module"),
> +            words("to get access to")] ++
> +                ConflictExportedPredPieces
> +    ),
> +
> +    MovingPieces =
> +        [words("The set of predicates and/or functions reachable from"),
> +        words("the proposed-to-be-moved predicates and/or functions,"),
> +        words("which should therefore be moved to the new module,"),
> +        words("would be")] ++
> +            MovingPredPieces,
> +
> +    ( if set_tree234.is_empty(MovingStayingPredIdSet) then
> +        MovingStayingPieces = []
> +    else
> +        MovingStayingPredPieces = name_set_to_line_pieces(ModuleInfo,
> +            MovingStayingPredIdSet),
> +        MovingStayingPieces =
> +            [words("However, the following local predicates and/or functions"),
> +            words("are reachable both from code being moved and"),
> +            words("code that is staying, which means that they would"),
> +            words("need to be either duplicated, or, if included in only"),
> +            words("one of the two modules, old and new, they would"),
> +            words("need to be exported from the module they end up in"),
> +            words("to be accessible from the other module."),
> +            words("Neither options is usually a good idea.")] ++

s/options/option/

That looks fine otherwise.

Julien.