On Mon, Dec 14, 1998 at 1:55 PM, Clyde wrote:
> On Mon, Dec 14, 1998 at 02:03:55PM -0800, Michael J. Ghormley wrote:
> > If I have inadvertatly allowed an INT on PortB or something, at
> least it has a
> > minimal effect.  I have always thought that this is a good idea.
>
> IMHO I think it's not such a good idea. What I would rather do is place
> some code at the interrupt vector location that did something noticeable
> then hung, e.g. sit in a tight loop toggling an output pin.

I agree with Clyde, this is another standard "good practice" for embedded
system developers, regardless of the processor.  The toggle-pin-til-watchdog
trick is also common for capturing fatal initialization errors, thus
allowing the developer to first test the debug pin if there's a system
that's reseting constantly or sporadically.  I tend to avoid this in the PIC
because I usually don't have a pin (or sometimes the few words of codespace)
available.

Instead, my startup code stores the reason for reset into a byte of RAM for
later use, if I want it.  If it's anything other than a regular power-up,
the reset reason gets spewed to the serial port, LCD, or even temporarily to
the LED display (if there is one).  This way watchdog timeouts, brown-outs,
and unexpected interrupts are all captured and dealt with the same way.

> Basically the philosophy is that there is no such thing as a small error
> in an embedded system

I also go by this philosophy, as should anyone developing commercial
products (personal Christmas presents exempt).  Ignoring things like
unexpected interrupts is often the equivalent to ignoring tips of icebergs.

-Ed V.
Agile Controls
evan@direct.ca
* Want to leave the piclist?  See
http://persweb.direct.ca/evan/pic/piclist.htm *