Re: [Gzz] Tuukka 2002-07-26 (languages, cleaning)

[b . fallenstein]

> Tuomas agreed we can backport the code to 2.1 so we wouldn't depend on
> Python at all. Later, if all Java code is rewritten in Python, we have the
> possibility of changing the interpreter from Jython to Python. Should we
> target that?

Too slow. Please, don't rewrite vobs in Python so that I have to run them in
Jython-on-Java-without-JIT (on my Mac: kaffe does *not* have JIT on ppc yet,
and IBM's new ppc JIT segfaults on me, but that's just BTW).

Python or Python plus C/C++ is worth thinking (I wanna run on pure Free
Software...), and it's a thought I've had, too, but rewriting *everything* in
Python while we're still running it on slow Java VMs is not gonna work.

We *could* try to target Java VMs and Python at the same time, providing
Python (or C/C++ extension modules) implementations of all interfaces. Tuomas is
worried about the client being in Jython for performance reasons I think,
but I also think that except for the startup that would be ok-- would have to
test.

Anyway, it's certainly not a short term goal IMO (short term goal is to get
0.6 out). For purely social reasons, I would also in any case ask very much
for keeping Java at least till end of 2002: The Gzz project at my school was
accepted on the base of it being a Java programming project (I know we haven't
contributed very much yet, but I still want it to go on).

- Benja