"DoN. Nichols" wrote in message
...
On 2013-07-28, Jim Wilkins wrote:

O.K. Yes, interpreted makes debugging and experimenting easier,
as long as the compiled code behaves the same as the interpreted
code. :-)

Functionally the execution appeared to be identical but of course the
timing was very different, so software-generated I2C communications
with the Device Under Test only worked properly when compiled. My
automatic timing calibration routine had a limited dynamic range that
couldn't handle both 2 GHz compiled and 75 MHz interpreted.

The target device was an early Power Over Ethernet controller that
could do some serious harm if it turned on the 48 VDC when it
shouldn't. I was skywiring those 0.5mm-pitch QFPs over the charred
crater where the engineer had burned one up.

It wasn't as bad as the Hot Swap PCI card controller whose
applications board shorted out the power supply on command. Hopefully
the IC limited the current well and fast enough to keep mobo voltages
stable. That was the most difficult circuit board outline I had to
construct, followed closely by the 3-dimensional Segway Balance Sensor
Assembly.

jsw