Detectin mainloop integration

[Havoc Pennington]

Hi,

On 9/20/07, Alexander Larsson <alexl at redhat.com> wrote:
> I don't really need my own integration, at least for the standandard
> connection that I use for async calls. However, there are some issues
> that made me do it:
>
> 1) dbus-glib is api unstable

I think a nice goal for the next few months would be to get at least
the main loop integration in some standard and stable spot.
Realistically, the "setup with main loop" call is not exactly going to
change API; it's one function.

> 2) dbus-glib also contains a huge chunk of other code (object mapping)
>   which I don't need. Especially at this low level in the stack.

Right, and in general I don't think the object system mapping is all
that "worked out" what it should be like. It would be good to detach
its API stability from just the simple main loop hook.

> 4) I looked at the dbus-glib mainloop integration code and it looked a
>   bit iffy. Unfortunately I don't remember the details...

In any case the API is fine, if it has bugs, it's not something that
should be hard to fix. A good bit of the desktop already relies on it
in any case.

Havoc