RE: Gtk/Gnome release schedules

[Kristian Rietveld <kris gtk org>]

On Thu, 2003-02-06 at 12:58, Michael Meeks wrote:
> Hi Martyn, Bob, Owen,
> 
> 	Since this is an emotive topic, and there are never any shortage of
> people crawling out of the woodwork to tell Owen to work even harder /
> faster :-) I think we should recognise, that since we are lumbered with
> at least 1 deprecated file selector (FS) API, we all agree that it's
> worth taking time to make sure that the new one is solid.

Indeed.

> 
> 	Also; some of the other bits in the new gtk+ plan, such as (eg.) the
> auto-completion stuff for GtkEntries is (perhaps) tied to the
> implementation of the FS.[1], and also fairly tricky to get right
> (AFAIR from talking to Clahey/Trow).

IMHO it might be nice if the fileselector used the new completion stuff.
I have been playing with completion during the combo development (as I
think completion is related to the TextCombo, as is History). Anyway, I
hope to finish combo RSN and send some info/thoughts/etc to the list
soon.

> 
> 	It seems to me that we should all wait until Owen is 100% happy with
> the new FS code, which could be as long as it takes.
> 
> 	Having said that - it would be nice to think that if
> Owen/Kris/Hadess/Whomever got abnormally hyperactive and produce
> something that everyone likes RSN, that we could (perhaps) do a stop-gap

Producing something that everyone likes RSN is something which we all
will regret in the end. Let's not go there.

> Gtk+-2.4 release, based off the 2.2 branch, with only the new FS API [
> and internal auto-completion stuff ? ].

I am wondering if releasing a new gtk+ every 5 or 6 months, like GNOME,
is a good idea. I would rather see releases of gtk+ every 9 or 12
months. Remember that there are also other projects outside of gnome
depending on GTK+. 

Personally I don't like releasing 2.4 as a 2.2 + new fileselector. 2.2
was 2.0 + multihead and font stuff (which I personally didn't like
either, but was needed). If we make 2.4 a fast release again, just like
2.2, we again won't have time to do other things which we have to get
done but take some time. As a result those things will be delayed for
sure, 2.2 also took more time than originally planned.

I am afraid that the same situation will exist with 2.6, "oh we really
need this in gtk+ RSN, please do 2.6 soon!", and I don't think we should
go there. In the end we will never get some important other things done.

And I am aware that this is a serious problem. Shortening the gtk+
release interval is not good, putting prototype APIs in a gnome release
is also bad. So the only options left are increasing the gnome release
interval, or put gtk+ 2.4 in gnome 2.6. Personally I would vote for
increasing the gnome release interval to something between 9 and 12
months and get gtk+ 2.4 in gnome 2.4 as a result.



	-Kris


> 	Owen - is that a reasonable straw for the hopeful to clutch at for the
> moment ? that _might_ then result in getting something beautiful for
> Gnome 2.4 - assuming that is that Gnome sticks with the 6 month
> schedule.
>
> 	Regards,
> 
> 		Michael.
> 
> [1] - that is assuming that Gtk+ will not provide a fully abstract file 
>       selection API, without wanting to provide an implementation too.