On 1/4/2013 1:53 PM, William Sommerwerck wrote:
"Jon Elson" wrote in message
...
William Sommerwerck wrote:

With respect to the American space program... While working
for RCA, I nearly destroyed one of the TIROS satellites. I don't
remember whether it was 6 or 7...

Well, that sounds like it could be one of those "tales from the cube"
stories on the back page of EDN. Tell us more, if you can!

Both satellites used dual, identical computers, with one backing up the
other. The software was periodically modified or updated, and I was
assigned to write the step-by-step instructions for a change.

Although my work was checked by other people, they made the same mistake
I did. I assumed that, because one of the satellites did not perform a
checksum on that section of program memory, neither did the other one.
But it did.

The result was that, shortly after the update, the checksum was
performed, wasn't correct, and the computer shut down. For reasons I
never understood, the backup computer also shut down. (This was an
intentional part of the design.)

I quickly figured out what was going on, and my boss complimented me
(which he shouldn't have -- he should have chewed me out). No one was
fired, or even reprimanded. I suspect this sort of mistake was not
uncommon.

The satellite was "fixed" by sending commands to reload and restart the
computers.

You never tested your code in a real setup?