>Have you tried making the interrupt service routine immediately turn on an
>LED, not even save any registers, to see if the interrupt is really getting
>called?

Solved it!  Your suggestion put me on the right track.

What was happening was - the interrupt happened only once and then never
again.  I'd correctly implemented the context saving bits of code - but
I'd forgotten to explicitly clear PCLATH,3 after saving PCLATH to a
temp_lath register.  So, if PCLATH,3 was set and an interrupt occured
(as it would when I was looping around in page 1), then the first GOTO
in the interrupt routine would jump to page 1 (shortly before the loop
as it happened!) and since the RETFIE never happened, the interrupt
never happens again, and it just loops indefinitely.

Easy in hindsight isn't it!!!
--
Jim