[Uim] announcement of SigScheme

[Kazuki Ohta]

Hi, Tokunaga-san.

I've already implemented the basic feature of R5RS except 
continuation and macro. Then, I want to step to the next 
stage, that is, check the all scheme code which assumes 
R4RS (SIOD's behavior). I think I should have the another 
branch for this work. Tokunaga-san, can I have the new 
branch?

# and also does uim's public svn access is not yet restored?

On Wednesday 13 July 2005 07:09, Kazuki Ohta wrote:
> Hi, All.
>
> I've uploaded the latest snapshot:-)
> URL :
> http://mover.cool.ne.jp/others/sigscheme/sigscheme-2005-0
>7-13.tar.gz
>
> I've introduced the simple benchmark and testing
> framework for this version by confirming its speed and
> compatibility to R5RS. Please try ./runbench.sh and
> ./runtest.sh.
>
> I use benchmarking programs listed in
> URL : http://www.cs.bath.ac.uk/~jap/ak1/youtoo/Bench/
>
> Roughly speeking, SigScheme wins SIOD in some benchmarks
> like bench-fib.scm even at this stage (of course, same
> heap size and same name hash size).  rec, tak, takl, takr
> also wins.
>
> # SigScheme has the tactics to get more faster and more
> # memory less.
>
> [ bench-fib.scm ]
> ubuntu% time ./sscm bench/bench-fib.scm
> 832040
> ./sscm bench/bench-fib.scm  1.67s user 0.03s system 98%
> cpu 1.727 total
> ubuntu% time uim-sh -B < bench/bench-fib.scm
> 832040
> uim-sh -B < bench/bench-fib.scm  2.27s user 0.01s system
> 98% cpu 2.310 total
>
> Otherwise, bench-arith.scm, bench-loop.scm loses.
> For bench-arithint.scm, our interpreter seems to scan the
> stack tremendous bytes. This may be the bug of GC.
> For bench-loop.scm, SigScheme needs much time for
> allocating and initializing the big size of heaps. I must
> introduce some methods to reduce startup overhead.
>
> regards.
>
> > Hi Kazuki, I'm looking forward to growth of SigScheme.
> >
> > At Sat, 9 Jul 2005 01:45:08 +0900,
> >
> > mover at hct.zaq.ne.jp wrote:
> > > URL :
> > > http://mover.cool.ne.jp/others/sigscheme/sigscheme-20
> > >05 -07-09.tar.gz
> >
> > There are no copyright information and license terms
> > (although your name is recorded as 'Author'). It's
> > serious problem and should be resolved ASAP. Your
> > rights are not protected!
> >
> > The lacking of them will bring unhappy troubles and
> > doubt such as stealing to uim project.
> >
> > I mean it.
> >
> > -------------------------------
> > YamaKen  yamaken at bp.iij4u.or.jp
> > _______________________________________________
> > uim mailing list
> > uim at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/uim

-- 
-------------------------------------------------
Kazuki Ohta : mover at hct.zaq.ne.jp
-------------------------------------------------