Oh well...

My point was that the carry bit has nothing to do
with the *acualy used* instruction in the code, rlncf.

In the lines after (that you snipped) I *did* say that
you instead has to deal with the MSB of the rotated byte.
(Which probably will be much less "unknown" then the C bit.)

I also showed the possibility of adding W with itself
on the 18-series, so you don't need the separate temp
variable (as in Alan's method). (I hope it works ! :-) )

Never mind...

Jan-Erik.




-----Original Message-----
From: Olin Lathrop [mailto:olin_piclist@EMBEDINC.COM]
Sent: den 16 oktober 2003 13:51
To: PICLIST@MITVMA.MIT.EDU
Subject: Re: [PIC:] RETLW on 18F


>> First doing a rotate can be a bad way to do it, because you first need
>> to clear the carry bit.
>
> Note the use of RLNCF "rotate left no carry".

That only means that the LSB comes from the MSB instead of the C flag.
You still have to deal with the unknown bit value.  Alan's method
essentially does a left shift of 1 and guarantees that the LSB will be
zero in one instruction.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.