Now I *am* truly pulling my hair out.  After finding
that the on-board oscillator on the 14000 varies considerably
from chip to chip, I rewrote my synchronous serial port code
to establish its timing from the on-board clock calibration
constant.  I've got the serial output going on on pion3 of PORT A,
and that's the only pin of PORTA I'm using.  The serial pin writes
to a 9600b serial LCD for debugging.

I wrote a function putch() to output a character, and putstr()
to dump a string.

It all works absolutely perfectly in the test main() I put
together, but when I plug the functions into my real application code,
it works 1 in 20 times (of applying power).  When it doesn't work,
the pin just goes low and stays there.

I'm definitely not tromping TRISA or anything in the rest
of the code, and I've been trying to fix this for three
days now...

-Will
"Life is a meritocracy"