[webkit-dev] Moving forward with WebKit/GTK+

[Gustavo Noronha Silva]

[this message is some days old, I'm having to send it now because my
SMTP was broken for some days]

Hello,

Sorry for the delay in responding, I left to Latinoware tuesday night,
and I didn't think I'd be with no Internet connection for so long.

On Wed, 2008-10-29 at 12:16 +0100, Holger Freyther wrote:
> The main difficulty besides lacking time is meeting the quality standards of the 
> WebKit project. Our WebKit/Gtk+ port is not using the LayoutTest regression 
> suite yet(I have started working on it, time is the limitation there), and we 
> have no GUnit coverage of the API.

That is quite true, I'm sure. But it's kind of a chicken and egg
problem: we need to go forward and keep/increase the developer mindshare
we've got to gain more and better contributors. If we look like we're
stagnant, it also hurts this goal.

> This makes it very hard to judge what is going to break. Besides that we 
> should probably have more Gtk+ reviewers as Alp and me sadly don't have the 
> time for WebKit that would be needed.

I didn't even know you were a reviewer, so forgive me for not mentioning
you. I have seen some of your posts in bug reports I read. But yeah, we
do need more reviewers in my opinion. 

Let me quote your other email here, too:

> The other part is, it is really hard to judge what is Gtk'ish. E.g. our 
> current settings API is using GObject properties. Now we have a patch to add a 
> thousand getters and setters as shortcut[1].

I agree, this is hard. This is one reason why I believe Christian Dywan
would make a good reviewer: he does have a very good knowledge of the
GTK+-api. Now, I usually don't think we need a lot of getters and
setters, but others parts of the API are already providing them.

I think we have time to make mistakes right now, though. Even if it
would be best to have a stable API from the begining, GTK+ is breaking
the API/ABI soon, and we can take advantage of that.

> I think this is weird, e.g. one setter/getter and enum would be enough, or 
> even the gobject properites are good enough... so some sort of API feedback of 
> key people would be good.

I agree. But we need to get those people into a mailing list, and invite
them to review API bits we are not sure about. I am pretty sure many
GNOME-related projects members would be willing to be involved in such a
mailing list =).

We can even request one at lists.gnome.org!

By the way, thanks a lot for reviewing/fixing/committing the
WebInspector patch! That's a personal itch, since I'm a web developer
and really want to stop using Firefox/Firebug =).

See you,

-- 
Gustavo Noronha Silva <gns at gnome.org>
GNOME contributor: http://www.gnome.org/