Re: gir stability

[Tomeu Vizoso <tomeu sugarlabs org>]

On Tue, Jan 31, 2012 at 21:37, Matthias Clasen
<matthias clasen gmail com> wrote:
> On Tue, Jan 31, 2012 at 2:40 PM, Ryan Lortie <desrt desrt ca> wrote:
>
>>
>> What makes this particular incident of note is that popup_for_device()
>> existed in Gtk 3.2 and was bound as such.  Python applications[1]
>> started using it, and then, with this commit, it disappeared.
>>
>> We need to figure out what our story is with respect to annotations.
>> 'Rename to:' is an extreme example (since an entire function, as named,
>> disappears) but we can easily cause problems just as serious with
>> changes that look a lot more innocent (like changing array length
>> parameters or such).  We can even break vala bindings with the
>> introduction of an (allow none).
>
> Thanks for raising the issue. In this particular case, the only reason
> that popup_for_device was added in the first place was to have a
> bindable version of popup - the 'rename to' annotation should really
> have been there from the beginning. But its a fair question to ask -
> at what point do we stop considering the introspected bindings a
> work-in-progress and treat annotations as equally part of the stable
> api ? Sounds like your take is that the answer has to be 'now'...

I'm afraid that if we consider introspected bindings stable right now,
we'll have to keep compatibility with lots of bugs, it will be really
ugly.

What if we mark classes stable one by one as people review them?
Ideally they would be reviewed by application authors depending on
what their applications use, so the most used would be reviewed first,
the more esoteric classes would be left for later.

Regards,

Tomeu