Re: [PATCH] C++: fix-it hints suggesting accessors for private fields

[Nathan Sidwell <nathan at acm dot org>]

On 04/25/2017 07:46 AM, Nathan Sidwell wrote:
> On 04/24/2017 04:06 PM, David Malcolm wrote:
>
> > test.cc:12:13: note: field ‘int foo::m_field’ can be accessed via ‘int 
> > foo::get_field() const’
> >     return f->m_field;
> >               ^~~~~~~
> >               get_field()
> >
> > Assuming that an IDE can offer to apply fix-it hints, this should
> > make it easier to handle refactorings where one makes a field
> > private and adds a getter.
> >
> > It also helps by letting the user know that a getter exists, and
> > the name of the getter ("is it "field", "get_field", etc?").
>
> Neat!
>
> > OK for trunk?
> >
> > gcc/cp/ChangeLog:
> >     * call.c (maybe_suggest_accessor): New function.
> >     (enforce_access): Call maybe_suggest_accessor for inaccessible
> >     decls.
> >     * cp-tree.h (locate_field_accessor): New decl.
> >     * search.c (matches_code_and_type_p): New function.
> >     (field_access_p): New function.
> >     (direct_accessor_p): New function.
> >     (reference_accessor_p): New function.
> >     (field_accessor_p): New function.
> >     (dfs_locate_field_accessor_pre): New function.
> >     (locate_field_accessor): New function.
>
> ok.

Oh, what if the field is being accessed for modification or lvalueness? 
Must the accessor return T, or can it return 'T cv &'?  I.e. does it 
need to look for setters too?

nathan
--
Nathan Sidwell