...seems to be bassackward, somehow. Did you mean to write:
"Most errors that *should* have been caught earlier aren't really errors"
instead of "shouldn't"?
If so, then this makes sense, even in the embedded context, where, "If it *can't* be 'fatal', then it *must* be 'caught'."; and if it *is* caught, it is not an error, but an anticipated failure mode of the system.
Handling everything that turns up in an FMEA[1], gracefully, is no trivial task--especially when the ESD protection on your microcontroller lets it run, with a supply rail *open*, on current drawn through input ports! [2] This is why I like embedded work: it does away neatly with the "ideal hardware" assumptions made in academia, and quickly separates the hard-boiled practitioners from the ivory-tower theoreticians.
---
[1] Failure Modes and Effects Analysis. One of the many things we enginerds are paid our so-called "big bucks" to do, is to determine, "How can it break?"
[2] Yes, it really happened, creating a critical safety issue that had been previously unanticipated. It was, shall we say, interesting to redesign the hardware and software to catch the problem, shut down the control, and register a fault requiring service.
--
HadEnuf?
|
