"Peter Duniho" wrote in message
...
"Tony Cox" wrote in message
.net...

You really don't need to take my word for it. Go see what NIST
has to say. http://www.nist.gov/dads/HTML/cyclic...ancyCheck.html

I saw what NIST has to say. So?

Then you'll have noticed the statement "Many transmission errors
may be detected, and some corrected" in their description of the
algorithm, right?

Just because someone wrote it, that doesn't make it true.

Indeed, but this is the National Institute of Standards and Technology,
not some random collection of net kooks. Say, why don't you write
them and tell them they are wrong, eh?


Well, here's a tutorial which claims (I've not followed their proof) to
show how to use CRC's to correct burst transmission errors.

Did you read the tutorial? I was unable to read it completely, because
all
of the embedded items are of a type that doesn't display on my computer.

I couldn't get it to display either, but I put that down to my ancient
web browser. Look, it's not my job to tutor people in coding
theory. Go search the web yourself if you're interested.

For a CRC to be useful in *correcting* erroneous data, it needs to contain
as much information as was lost in the first place. In the example you're
talking about, where the error is limited to a certain area of the data,
you'll find that the CRC itself contains essentially the same information
that was lost.

No ****, Einstein. Say, you don't suppose that might be how
that corrupted data can get reconstructed do you?

But wait. You started by vehemently denying that CRC's
provided error correction; now you're saying they can. Guess
you must get a buzz from pointlessly arguing with people. Sorry
Pete, but it doesn't do a thing for me so I'll stop..