> Does anyone know if the previous problems with that have been fixed?
It's not possible to fix without a significant redesign of the port
logic..

Does
> anyone use the port change interrupt while they are changing the other bits
> on that port?  Or is the port change interrupt still only useful for waking
> the chip from sleep?

It's useful, as long as you don't try to manipulate ANY other
bits on the port.
ALL instructions are read/modify/write so any attempt to change
bits on the port causes a read, and that resets the bit change
registers, so you can miss the change if it happens right
at the moment you access the port. It's a fundamental design
flaw of the port. You can use it for a SINGLE bit, otherwise
you risk missing an edge (since a read of the port to discover
which bit changed also resets the detect register).

I've been using PORTB change on a 16F876 to capture the
pulse width of an ADXL202. I found that I was missing about
1% of the edges because I also had to bang other port
bits for other functions. I simply throw away abnormal values
(since they're out by a factor of 2) and the sampling rate was
not critical.

Robert

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