Re: Deja Vu: multiple definition of `java::lang::String::length()'?

[Adam Megacz <gcj at lists dot megacz dot com>]

Bryce McKinlay <bryce@waitaki.otago.ac.nz> writes:
> > Oh how I wish you weren't kidding =)

> Who says I'm kidding! There are so many reasons why this would make 
> sense.

Rock on, this sounds like fun. Let me know if/when you start on it,
I'd like to help. Especially if it involves creating a cleaner OO
equivalent for GCC trees, and transforming them into 'real' gcc trees
as the last step before handing them off to the backend.

If we had something like that, I'd probably write some toy functional
language front ends -- I'd really like to do that, but not enough to
justify any more struggling with gcc's trees ;)

Another cool thing you could do is limit the expressive power of the
Java trees to exclude things that can't be expressed in bytecode
(pointer arithmetic, etc) and have 'gcj -C' transform Java trees into
bytecode. Then if you write frontends for other safe languages
(Haskell, ML, Clean, etc) using the Java trees, you would get both
compile-to-bytecode and compile-to-native with no additional work.

Okay, I'm going to stop dreaming now...

  - a

-- 
"If I put copyrighted material into the 'chewy nougat center', I can
 claim the crunchy chocolate coating is an 'Access Control
 Mechanism'."                                     --lynx_user_abroad