Re: [Vala] A struct in a vapi file with undefined contents?

[Dylan McCall <dylanmccall gmail com>]

And, naturally, if I'd just not made a silly typo when I had tried
this, I would have seen valac doesn't complain at all and it works the
way I want, allocating the correct space in memory. Hooray!

So, I have the following in my vapi file now…

[CCode (cname = "FIT_CONVERT_STATE")]
public struct State {
// Private
}

Sorry about the noise. Of course, please do point out if this is going
to end poorly :)

--
Dylan

On 28 July 2013 22:28, Dylan McCall <dylanmccall gmail com> wrote:
> Sometimes, when I find myself working with a (plain) C library through
> Vala, I end up using a pattern like this. I really don't like it:
>
> void * convert_state = Fit.create_convert_state();
> Fit.Convert.init(convert_state, true);
>
> In this case, Fit.Convert.init is part of the original library, which
> I made a vapi file for. It takes a reference to convert_state, which
> it writes to. Fit.create_convert_state() is a function I created in a
> small C wrapper library. The entire point of it is to allocate space
> for a struct on the heap (for Fit.Convert.init), and then return a
> pointer to that struct. The struct isn't in the vapi file because it's
> huge and complicated and it's really internal to that library. Its
> contents aren't _interesting_ to the Vala program - just that it's of
> a particular size.
>
> Is there a way to add a struct with undefined contents in a vapi file?
> That way I can maybe do something like this with the example?
>
> ConvertState convert_state = ConvertState();
> Fit.Convert.init(convert_state, true);
>
> We wouldn't be able to access any members in convert_state, but we'd
> be spared ugly void pointers.
>
> …or am I looking at this the wrong way? I probably am. Any help is
> appreciated :)
>
> --
> Dylan