On Feb 29, 10:16=A0am, Roy <roy.van.r...@[EMAIL PROTECTED]
> wrote:
> ...
> But there is just a little bug now, you jump to qPtr instead of the
> qDec (the decoder)...!!
Oops, made a global replacement at the last minute.
There are six 1- and twelve 2-jmp possibilities. 18*4 =3D 72 scans,
I've only been going for 40! If I set my generator to use all 18,
then I would have set of values for 72 scans that could be trimmed
_down_ instead of _up_ :-)
1-jmps:
_no mutation_
{
}
<
>
djn.f
2-jmps:
{{
{}
{>
{djn.f
}<
}>
}djn.f
<<
<>
<djn.f
>>
>djn.f
> ...
> A good thing indeed is to put all the direct jumps to qDec on top, the
slo=
wer
> scans to the bottom, got me some extra points.
Good idea :-)
> >
> > Even more dense code might look like:
> > =A0 =A0seq =A0A,B
> > =A0 =A0jmn.f dec,C
>
> > Three scans in two lines, but I don't see how it can be made to work.
>
> I think thats impossible. The problem is the first scan, you can't
> really change the flow it follows.. But who knows, maybe some clever
> redirection will make it possible!
Maybe just one seq/jmn.f pair could be squeezed in.
P. Kline
|
