Ok for a 0.24 release?

[Havoc Pennington]

On Mon, 2005-01-31 at 14:27 +0100, Benedikt Meurer wrote:
> Havoc Pennington wrote:
> > davidz suggests that we go to 0.30 to help make visible the large
> > breakage.
> 
> How about maintaining backward-compatibility for 1 or 2 releases? I
> would really appreciate that, and I bet packagers and other developers
> will appreciate that as well, as it allows for a smooth upgrade rather
> that a large breakage.
> 

We aren't there yet, as the README explains. When we get to 1.0 there
will be a strict and rigorous back compat policy on both library and
protocol. Until we get to 1.0, there are things that need to change in
order to move us toward 1.0. If we leave it compatible too often it will
eventually create a "de facto" 1.0 where breaking it is almost
impossible and I very much want to avoid that...

In any case, we are close to 1.0 and the steps are documented in
doc/TODO; this isn't a blue sky or mysterious milestone. It just needs
work. Right now it's proceeding at a slow pace because it's mostly just
me doing it. I have been prioritizing the breakage-causing work because
I want to be testing the final intended APIs, not APIs we know we want
to change.

FWIW there are no "major breakage" items left in libdbus proper in
doc/TODO, with the possible exception of rearranging dbus-bus.h. Two
more breakage areas are expected though: 1) glib bindings as we figure
out the right way to implement them and 2) in response to real-world
experience as we start using dbus within the desktop projects.

The goal here is to avoid locking it down while fatal flaws still exist.
We all have working IPC systems from the past, there is no point
replacing them with something broken.

Havoc