Some time ago I wrote:

>
> > I wanted to implement a pseudo-op called 'J' which used the shortest
> > possible opcode JR or JP.  It turned out that to optimally implement
> > this, the assembler would potentially have to make a large number of
> > passes, refining its estimate of the value of each label each time.  It
> > is also conceivable that the assembler could get stuck in an infinite
> > loop.
>

Thanks to all the folks who sent algorithms etc. for solving this
problem.  Actually, I realised that there was no necessity for an
'infinite loop' to occur with a proper algorithm.  Rather, I meant
this was a possibility if due care was not exercised.  Since this
project was done in a great hurry (just a few days!) I didn't have
time to consider such luxuries.  If I was going to do another
assembler, I might consider a PIC assembler which would indeed have
this sort of feature so that opcodes could be conditionally
inserted without breaking the assembler.

Tell you what, though, there's nothing like writing an assembler
and simulator to teach one the instruction set in a hurry!

Regards,
SJH
Canberra, Australia