Ronald Bultje wrote:
> >
> > Gnome and KDE sit on my harddrive fine, and never bite
> > each other. Sometimes they even do coffee, though not
> > in public.
>
> There's a difference here. Suppose you have a commercial application.
> Say, opera. I can now decide to use eithet Gtk+/Gnome or Qt/KDE.
> Whichever I choose, I know that it will work on the clients desktop,
> that's all I care about. So I can choose... Whichever I choose, everyone
> can run it. So, no problem.
Ah, but my Flash plug-in doesn't work with Opera,
so I have to use Netscape, so I'm back to square one.
> Now imagine that I fork video4linux, so I have a video4linux interface
> 'video4linuxGerd' and 'video4linuxRonald', which are incompatible, just
> like Gtk+/Qt are incompatible. Now commercial company Billacle wants to
> make linux drivers for their winVT card. They know that they have to
> choose between either of the one interfaces. Making a driver for both
> interfaces sucks major ass, just like making opera for Gtk+ and for Qt
> is a bit overdone. So, they need to choose. But they know that if they
> make a driver for video4linuxGerd, it will only work with xawtv, and if
> they make a vdeo4linuxRonald driver, it will only work with my
> application. This sucks, because now I don't have a standard anymore,
> and in this case I need one. I have to choose and it won't work for
> everyone, it will only work for some people. Of course I can make
> convertors, or support both, but that's not my point.
All you're pointing out is that life sucks, because you have to
choose between compatibility and standardization in varying amounts.
Even if you get a standard multimedia architecture for Linux,
will it be standard for xBSD, Sun and MacOSX? Vendors will still have
to decide what combinations, if any, to support.
Now, if you can get projects to agree on some common ground
in a reasonable way that makes them happy, fine. But concluding
one way is the right way and needs to be standardized on is
the subject of the famous Tanenbaum ("the right way") vs.
Linux ("the arrogant upstart") conversation. But history is
full of brilliant deadends. People seem enthusiastic about
Broadcast2000, but it seems dead as a serious project.
GGI seemed like a great idea that just ran out of gas.
As someone noted, work on what feels good, and let the market
pick a (some) de facto standard(s).
http://www.oreilly.com/catalog/opensources/book/appa.html
[Note: I fear we're losing V4L relevance here]
--
Bill Eldridge
Radio Free Asia
bill@xxxxxxx
