"I wasn't necessarily saying Vala was an app development language (although it could be)."
GNOME devs are already writing full apps with it, so it is being used as such.
"If Gtk used some other language like, say, Python, with Python GTK widgets and so forth, then how would Java, C#, or even C or C++ have access to these components?"
yes, i understand that is what has driven Vala's design and implementation. i just don't agree with the "we'll wedge another home grown language in between the C and the other languages" approach. i think it's overly complicated and limits the number of people who can (and will) hack on it.
time will tell if i'm smoking crack or not, of course .. =)
"Vala isn't actually a language in the same sense as Python, C# or Java is. It's really just a syntactic extension of C and produces plain, simple, C code."
that is produces C is both a feature and a bug. it makes debugging much more awkward (and for a while wasn't even possible at all! how do you go from your generated C to your Vala code in gdb? there's a plugin now for gdb, but really .. oy vey!) and you lose all the interesting security possibilities of managed code.
which is precisely my point.
"it will be possible to extend and improve GTK to equal Qt while still maintaining the ability to use it from any language binding."
let's do this then: let's come back to this in 2-3 years (it takes time to get these things going, i know) and see if that theory works out.
my theory is that it will just be one more baroque tool that people working with Gtk will have to get their head around (and people complained about moc with Qt; they ain't seen nothing yet ;) thereby limiting the pool of candidate developers. as a non-transferable skill it won't gain much in the way of value that might cause people to learn it "just because", and yet people will write applications with it. i expect to see more and more vala usage in Gtk+/GNOME (because, well, that's already happening =) and it will cause the project to become more insular rather than less.
i do expect that those using it will get more done with vala than with plain C, but not to an extent that will make up for the number of people lost by not choosing a language syntax that is already widely known or a language that avoids compile cycles, dealing with the intricacies of C debuging, etc.
given that it's homegrown, it will also soak up resources maintaining and extending vala itself that could be put elsewhere.
combined, i expect individual efficiency of existing contributors to increase due to vala, but the overall effect on the project to be a net negative. i predict that in a few years vala will get quietly binned. bonobo 2.0 if you will: a cool idea that "just has to work, it's so well designed and advanced!" but which just didn't pan out in reality.
again, i could be wrong. and i certainly don't want to see the GNOME team falter. but vala gives me the heebie-jeebies.