From: Godmar Back (gback@cs.utah.edu)
Date: Wed Jan 13 1999 - 14:06:52 EST
>
> > > May be this could be done by lt_dlopen() directly ?
> >
> > I'll commit this change in the next few seconds.
> I'll go home in the next few seconds :)
>
Well, tomorrow will be another exciting day of ignoring your regular
work and doing some kaffe hacking. Or do you get paid for doing kaffe
hacking?
I at least have the alibi (which actually isn't that much of an alibi)
that I need kaffe for my research.
Which gets me to another topic I'd like to bring up:
Why don't we shoot for the next release about 1-2 weeks after Peter
merges in his new awt? Also, let's prepare this next release a bit
better than the last one. What I mean is having a short, informal
code freeze period where only bugfixes go in.
I hope to have the gc subsystem untangled by then. And maybe Jason
will be done with his new mark-stack based gc. Then we should be able
to beat Sun's 1.1.7 without JIT for regular applications speedwise.
Plus, I hope to have the threading system untangled and maybe I get
the SMP port for linuxthreads working. Since linuxthreads are so
grossly inefficient, this would only make sense on a machine with more
than one processor. As luck will have it, they've just equipped our offices
with 21 dualprocessor machines last year. (Somewhat unfortunate, though,
only one runs Linux cause people here are FreeBSD aficionados. I'm
even worse: I run NT most of the time on my desk.)
- Godmar
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:41 EDT