Re: [Vala] Possible bug in naming conflicts?

["Michael B. Trausch" <mike trausch us>]

On Wed, 18 Feb 2009 10:34:34 +0100
Andrea Bolognani <eof kiyuko org> wrote:
> I agree with you that's kinda strange when you look at it from the
> Vala perspective, but it's standard in the GObject world to have a
> get_type() method which returns the GType associated with a class.
>
> I see three possible ways to resolve the problem:
>
>     1. Avoid the name clash by renaming the *user defined* get_type()
>        method to get_type_user() or something like that.
>
>        This would be awkward if the library is used from C, so I think
>        it's not really an option.
>
>     2. Leave the user defined method as it is, and rename the standard
>        get_type() to get_gtype().
>
>        This would break the GObject conventions; OTOH, the get_type()
>        function is usually called by a macro, so it shouldn't break
>        anything except maybe for users expectation.
>
>     3. Special-case the get_type() method, along with other methods
> that could cause clashes, and emit a nice error from valac, instead of
>        failing misteriously when gcc is run.
>
> The third one seems the best one to me; obviously, it's Jürg's call.

When stacking the three together, I have to say that I agree.  If the
goal is to retain 100% adherence to existing convention, the only
reasonable thing is to throw an error.  C is held together by
convention it would seem, and so it wouldn't appear to be a good idea
to disturb that.

Some sort of predictable name-mangling would also be possible, though
there'd of course have to be a convention for doing it.  :-)

        --- Mike

-- 
My sigfile ran away and is on hiatus.
http://www.trausch.us/