FW: gnu.bytecode & gnu.expr License
Mark S. Miller
Mon, 25 Oct 1999 18:27:43 -0700
At 06:01 PM 10/25/99 , Per Bothner wrote:
>I think the thing to do is for someone who knows E and is familar with
>Scheme to write up a design for how E constructs would be mapped into
>Scheme constructs, and then discuss likely problems. The we could
>think about how difficult if would be to surmount them. It's probably
>best that I should just read through the E Kernel Language
>specification; that would probably give me a good idea of the issues.
That sounds great. I'd been fantasizing for a while about defining the semantics of Kernel-E by showing a translation to Scheme. I think it should be straightforward, but please ask as soon as anything needs clarification. The current Kernel-E documentation is far from debugged.
Besides mapping Kernel-E to Scheme, there's the matter of E's inter-callability with Java. I suspect E does something different from Kawa here, and I would be surprised if this issue is orthogonal from code generation issues. Just something to keep in mind. Your proposal is the right thing to investigate first.
>No, but I have been very open to accepting changes that people need.
>You're not the first ones using gnu.bytecode and other Kawa-based
>packages. I am very interested in extending the code to work for
>other languages, and I generally accept clean patches. (I do
>recommend that people discuss the design for any major patches
>*before* serious coding, because I may have particular ideas about how
>to solve a problem.)
>Thus once I accept a patch and check it into the Kawa cvs repository,
>you're basically back to using an unmodified version.
> > ...
>I didn't expect you to. All I require is that any changes you make,
>you also submit to me (any time before you make a "release"), and
>work with me to make them into a form aceptable to both groups.
This seems pretty good, but I'm concerned about the time delay. Even with the best of intentions, we are all busy with many things, and some get postponed. I would hate to be stuck not being able to release a new E until the next release of Kawa comes out. If we can find an arrangement that alleviates the sequential dependency, then, given your response, I'm happy (and eager!) to move forward.