Alan Shoemaker wrote:
> Alan Shoemaker wrote:
> [snip]
> >
> > Finally I attempted an Air (7.0) install on my wife's machine
> > over her Venus (6.0) installation. It failed at the point of
> > initializing the cdrom. I made a boot disc to use the text
> > install and it failed too, at initializing the cdrom. I then
> > installed Helios (6.1) on that machine (to see if it would have
> > any problems) without a hitch.
> >
> [snip]
>
> This is an update on the above problem. I've successfully
> installed Air (7.0) on the above machine by swapping out the
> cdrom drive for another. After the drive change the
> installation went flawlessly. I guess that the drive was not
> reading the cd very well and at a certain point the read
> problems would crash the install. I've seen this behavior
> before, but still don't understand why it doesn't generate an
> error message?
Depends on the severity of the error. In algebraic coding theory we
can use the (much more complex) polynomial multiplication/division
schemes to recognize and correct m errors in a message of n bits
(utilizing primitive monic polynomials in the Galois field of
2^(m+n) for example). What happens with exactly one too many
errors? The error is recognized but not corrected (decoding
failure) and with more than that the message is decoded into the
WRONG one (decoding error).
The scheme used on CDs (and most use CRC or checksum) break down
when enough errors occur to introduce a correct checksum. This
means that the error occurs and is not recognized. Presto, a jmp
becomes an add and awayyyyy we go. In most cases, I have seen an
error message, but I have had The same result and usually changing
the CD drive (or just burning a copy from a drive that reads the CD
correctly) cures the problem.
The idea that CD drives are different enough to read badly kinda
gives me a warm and fuzzy feeling inside. I know I should have
gotten that flu shot.
Civileme
>
>
> Alan
--
experimentation involving more than 500 trials with an
ordinary slice of bread and a tablespoon of peanut butter
has determined that the probability a random toss will
land sticky side down (SSD) is approximately .98
