Using gcj as an extension language

Thu Dec 19 11:37:00 GMT 2002

>>>> The real question is what we need to ship to be
>>>> compatible with Sun's JDK.
>>>> It isn't that simple. First, there are applications out there using
>>>> AWT. I've been to meeting where lack of AWT immediately kills any
>>>> chance of using gcj. So, while I agree that AWT isn't a great
>>>> toolkit, nevertheless we need it as long as Sun ships it.

GNU usually works differently.
"GNU is not Unix" sums it up. Let the competition trademark their trade
names, it doesn't matter. Linux, as a trade name, has won or is about to win
from Unix. By staying compatible? Certainly not. It is now the Unices that
are trying to be compatible with Linux :)
Look at Richard Stallman. Regardless of what his opponents may say about him
and his outspoken and uncompromising attitude, his GPL is slowly, but surely
taking, over the entire scene. He's poised for getting a page in the history
books.
Trying to be compatible with the competition, gives the competition the
first-mover advantage. It doesn't take them much effort to break your
implementation. And you can play catch-up forever, from breakage to
breakage. It's hopeless. You actually do it unto yourself, by implicitly
promising your users that you will one day be compatible with Sun, even if
it doesn't really depend on you, but on Sun.
I would drop this unrealistic promise and resulting expectations, and become
instead a project with its own vision. There are so many things that Sun
does wrong and that GCJ can do right. No JRE. Small executables. Native
performance. Fast and feasible GUIs (with SWT). GCJ has the opportunity to
become "C++ done right".