Wine, WIDE & Unix (was: Support for pkgconfig)

[Mike Hearn]

> Really? Do people think that KDE & QT would be a more suitable solution?

No, but they are not the only platform available, as you have noted.

> platform we want to build? You can develop apps for free on Windows 
> (see MinGW, Borland, OpenWatcom), or for a small price buy MSVC, yet 

How many people actually program Windows without using an IDE? Most
programmers I know use either MS Visual C++/Basic or Delphi (and now
.NET).

> So the _only_ viable Unix GUI API is GTK/GNOME. 

I would say, rather,
"So the _only_ viable Unix GUI API for non-GPLd software is GTK/GNOME"

For GPLd software, Qt is fine (though aesthetically i prefer gtk). This
would only be a problem if in future you were required to use it in
order to get integration, as you are required to use the Win32 API
implementations to get integration on Windows. Luckily that isn't the
case, and is becoming less so every day.

I don't see any fundamental reason why you can't have the same app
written using Qt and GTK next to each other, with identical L&F and
perfect integration.  Sure, that's not the case today, but it's
possible.

> But version 2 of that API 
> came out only a year ago, and Evolution is not even ported to it! 

Well, it almost is. Apparently the 1.3.x series is quite stable now.

> Do you
> consider that traditional? It's so new it's not even funny. And maybe
> that's fine, but the problem is that they lost steam: at the moment it
> seems that most development is done by RedHat, Ximian, and Sun. 

Not sure where you got that idea, Ximian haven't been contributing much
code of late at all, they mainly maintain current stuff. Redhat and Sun
do quite a bit, but the bulk of the work, as ever, is done by
volunteers. The release coordinator is a volunteer, GStreamer is
developed by volunteers, a lot of the work on GTK is done by volunteers,
ditto for gnome-panel, libgnome and so on.

> And it's going slow. The two platforms (GNOME and KDE) seem to diverge more and
> more, and if you look asymptotically, do you see a bright, integrated
> future?

Actually.... yes? :) They aren't really diverging in my eyes, this
broken system tray I have sitting here taunting me is a visible reminder
that we need to keep Wine up to date with the latest standards.

> I don't think that WIDE is the greatest idea since slice bread. It's just
> something I thought of as a possibility during the discussion, and I threw
> it in there as a way to show there are other ways of looking at Wine.
> But I also don't think it's such a bad idea either. And I really don't
> understand the irrational fear (oh, it comes from MS, it *must* be evil),
> and the overly emotional and unsupportable defense of 'Unix API'.

I'm still not sure why you think we need one implementation to get
integration, as opposed to one set of interfaces with several possible
implementations. Experiments like the system tray seem to indicate that
we can get a unified platform between several large toolkit
implementations regardless of their license, or who designed them.

-- 
Mike Hearn <m.hearn at signal.qinetiq.com>
QinetiQ - Malvern Technology Center