[LINK] Spread of buggy software raises new questions

[Chirgwin, Richard]

Glen Turner wrote:

> As a test of this, draw a state machine of a washing machine.
> Now add in a "lid raised" event that should halt the wash
> and spin cycles and resume them when the lid is lowered.
> You'll get a difficult-to-verify mess.

Well, yes - but you only have to design it once. Once you have a state
machine which handles interrupt events (and I guess terminate events), you
put that in the PGA design library and drop it onto whatever bit of
programmable silicon you are using this time.

> As for your second question, you want microprocessors
> in appliances so that they are easier to use and cheaper
> to run.  

Except that "easier to use" includes the unwritten caveat "if it works".
This story was about it not working, partly for the usual reason given by
software developers - "but you can't test every line of code!". But are they
successfully easy to use? You give a good example, the FP washer. 

OTOH, from where I sit I can see a microwave which is very, very smart, but
you can't ask it for ten seconds to revive a cooled-off coffee. So you're
reduced to living within the parameters of what the developer though of. A
dumb device is more demanding on the user, but paradoxically it can do more
stuff because of the user input.

Richard C

PS, Thinks...Can't remember the correct symbols or sequences for a state
machine, but:

Physical inputs: 

Lid switch    \
Fill Alarm     | "OR" Gate
Balance Alarm /

"OR Gate" applied as a start/stop to the timing oscillator (from the old,
old days, a 555 timer!).

Maybe it's not such an impossible mess after all?
RC